fix(git-wrappers): no --login must use the caller's own credential, not a guessed shared login (#1352)

Co-authored-by: fargo <[email protected]>

.mosaic/repo.json

@@ -0,0 +1,4 @@
+{
+  "integration_trunk": "next",
+  "release_branch": "main"
+}

.woodpecker/ci-image.yml

@@ -22,9 +22,9 @@ steps:
     image: gcr.io/kaniko-project/executor:debug
     environment:
       REGISTRY_USER:
-        from_secret: gitea_username
+        from_secret: REGISTRY_USERNAME
       REGISTRY_PASS:
-        from_secret: gitea_password
+        from_secret: REGISTRY_PASSWORD
       CI_COMMIT_BRANCH: ${CI_COMMIT_BRANCH}
       CI_COMMIT_TAG: ${CI_COMMIT_TAG}
       CI_COMMIT_SHA: ${CI_COMMIT_SHA}

.woodpecker/ci.yml

@@ -2,16 +2,38 @@
 # node:24-alpine + python3/make/g++/postgresql-client + pnpm + a warm pnpm
 # store. The install step resolves from the baked store (--prefer-offline)
 # instead of paying a ~731s cold fetch + native compile every run.
+#
+# PINNED to an immutable lock-tag (#1328, brain D27): ci-image.yml pushes
+# lock-<sha256(pnpm-lock.yaml)[:12]> atomically with :latest, so the two are
+# byte-identical at push time. A mutable :latest resolves per-pod at pull time
+# on the k8s backend, which made CI verdicts non-reproducible (same tree, same
+# config, different images across runs; see #1324 comment 23382/23386). The pin
+# changes ONLY through reviewed commits; a wrong tag fails loudly at image pull.
+#
+# Bump procedure: when a recipe change (pnpm-lock.yaml / Dockerfile.ci) lands on
+# main, ci-image.yml pushes lock-<new>; a follow-up PR updates this anchor.
+# Until then pipelines keep the old pin: reproducible, with the documented
+# network-fallback lag (frozen-lockfile resolves missing packages from network).
+# Known limitation: lock- addresses the lockfile only, so a Dockerfile-only
+# change re-pushes the same tag with new content (#1328 follow-up: recipe-hash).
 variables:
-  - &node_image 'git.mosaicstack.dev/mosaicstack/stack/ci-base:latest'
+  - &node_image 'git.mosaicstack.dev/mosaicstack/stack/ci-base:lock-9cb7ffcd8828'
   - &enable_pnpm 'corepack enable'
 
 when:
-  # PR + manual CI run on any branch — the pull_request pipeline is the merge gate.
-  # push CI is restricted to protected branches (main) so a feature-branch push no
-  # longer fires a redundant SECOND pipeline alongside its PR pipeline. This ~halves
-  # CI load on the storage-constrained runner with zero loss of gating (branch
-  # protection requires no push/ci status context; main still gets full push CI).
+  # PR + manual CI run on any branch: the pull_request pipeline is the merge
+  # gate (next is protected and the default branch since 2026-08-19).
+  # Push CI runs on main only. next deliberately runs NO push ci: post-merge
+  # verification on next is carried by publish.yml's `verify` step
+  # (pnpm verify:release), which mirrors this pipeline's complete mandatory
+  # set step-for-step, enforced by scripts/verify-release.test.mjs. PR CI
+  # tests the PR HEAD tree (refs/pull/N/head, measured 2026-08-19), not a
+  # merge ref, so if next advances before a merge the landed tree differs
+  # from the tested one; publish verify re-runs the full set on the landed
+  # tree (PGlite path). Measured 2026-08-19: the 21 most recent push events
+  # on next each ran exactly one pipeline (publish), zero ci.
+  # Keeping push ci off next also avoids a redundant second full-suite run
+  # per merge on the storage-constrained runner.
   - event: [pull_request, manual]
   - event: push
     branch: main
@@ -30,6 +52,19 @@ steps:
       # the baked pnpm store.
       - pnpm install --frozen-lockfile --prefer-offline
 
+  # ---------------------------------------------------------------------------
+  # The steps below (sanitization, upgrade-guard, typecheck, lint, format,
+  # test) are the COMPLETE mandatory verification set. SDLC-D-034 mirrors them
+  # one-for-one in the canonical terminal verification command — root
+  # `pnpm verify:release` (scripts/verify-release.mjs) — which the publish
+  # pipeline (.woodpecker/publish.yml `verify` step) runs before ANY publish
+  # effect. These lines stay direct (not routed through the runner) because the
+  # #1017 test-enumeration guard audits framework tool paths through THIS
+  # surface; scripts/verify-release.test.mjs enforces that the runner's stage
+  # table keeps matching these commands exactly, so the two cannot drift.
+  # ---------------------------------------------------------------------------
+
+  # Canonical verify:release stage `sanitization`.
   # Blocking gate: public framework package must contain no operator-specific
   # personal data or private $HOME defaults. Runs early (no node_modules needed).
   sanitization:
@@ -69,6 +104,7 @@ steps:
       # stub supplies the scale instead of the host's own checkout.
       - bash packages/mosaic/framework/tools/git/test-mosaic-worktree-large-repo.sh
 
+  # Canonical verify:release stage `upgrade-guard`.
   # Blocking gate (#791): a framework upgrade must never write or delete an
   # operator-owned path. The HARD GATE proves an unanticipated operator sentinel
   # survives a keep-mode reseed byte-identical (with rsync present AND absent —
@@ -90,6 +126,8 @@ steps:
       - bash packages/mosaic/framework/tools/quality/scripts/test-upgrade-durable-snapshot.sh
       - bash packages/mosaic/framework/tools/quality/scripts/test-install-migration.sh
 
+  # Canonical verify:release stage `typecheck` — the same `pnpm typecheck`
+  # invocation (which runs the checkout preflight first, then turbo).
   typecheck:
     image: *node_image
     commands:
@@ -100,7 +138,8 @@ steps:
       - sanitization
       - upgrade-guard
 
-  # lint, format, and test are independent — run in parallel after typecheck
+  # lint, format, and test are independent — run in parallel after typecheck.
+  # Each runs exactly its canonical verify:release stage command.
   lint:
     image: *node_image
     commands:
@@ -117,6 +156,12 @@ steps:
     depends_on:
       - typecheck
 
+  # Canonical verify:release stage `test` — the `pnpm test` line below is the
+  # shared command; everything else in this step is PIPELINE-LEVEL
+  # prerequisite the canonical command expects its caller to provide (SDLC-D-034):
+  # the ci-postgres service + pg_isready wait + db:migrate (postgres path),
+  # `apk add openssl`, and the pinned pi install. None of those can move into
+  # the runner (it must also work locally on the PGlite path with no database).
   test:
     image: *node_image
     environment:

.woodpecker/publish.yml

@@ -1,10 +1,29 @@
 # Build, publish npm packages, and push Docker images
 # Runs on main for stable publishes and on next for integration-line prereleases/images
+#
+# SDLC-D-034 publish gate: every publish effect (publish-npm, publish-next-npm,
+# and every image build/push step) depends DIRECTLY on the `verify` step below.
+# `verify` (a) asserts the provider's commit identity matches the actual
+# checkout (CI_COMMIT_SHA == git rev-parse HEAD, fail closed on mismatch or
+# emptiness) and (b) runs the canonical terminal verification command
+# (`pnpm verify:release`), which mirrors the PR CI pipeline's complete
+# mandatory set (sanitization, upgrade-guard, preflight+typecheck, lint,
+# format:check, test, build) — see scripts/verify-release.mjs. A missing,
+# failed, skipped, cancelled, or inconclusive verification therefore skips the
+# dependent publish effects (fail closed). Path-filtered short-circuits may
+# skip publish EFFECTS (e.g. docs-only merges) but never bypass `verify` for a
+# publish that does run: `verify` itself carries no path filter.
+# scripts/verify-release.test.mjs enforces this DAG invariant at checkout time.
 
 variables:
   # Pre-baked CI base (see .woodpecker/ci-image.yml): node:24-alpine +
   # toolchain + warm pnpm store. Kills the second cold install publish pays.
-  - &node_image 'git.mosaicstack.dev/mosaicstack/stack/ci-base:latest'
+  # PINNED to the immutable lock-tag, not :latest (#1328, brain D27): a mutable
+  # tag resolves per-pod at pull time on the k8s backend and made CI verdicts
+  # non-reproducible (#1324). Byte-identical to :latest at pin time (pushed
+  # atomically by the same kaniko run, main 712c770, 2026-07-26). Bump only via
+  # reviewed PR, per the procedure in .woodpecker/ci.yml's header comment.
+  - &node_image 'git.mosaicstack.dev/mosaicstack/stack/ci-base:lock-9cb7ffcd8828'
   - &enable_pnpm 'corepack enable'
   # Heavy kaniko image builds (~25 min) — gate them so a merge that only touches
   # the npm-only CLI (@mosaicstack/mosaic) or docs does NOT rebuild the platform
@@ -48,6 +67,45 @@ steps:
       # Resolve from the baked pnpm store instead of a cold network fetch.
       - pnpm install --frozen-lockfile --prefer-offline
 
+  # SDLC-D-034 exact-commit publish gate. No `when`/path filter on purpose: it
+  # runs for every event this pipeline serves so no publish effect can ever
+  # start without it. Fails closed on commit-identity mismatch (or either SHA
+  # being empty) and on any incomplete verification.
+  verify:
+    image: *node_image
+    commands:
+      - *enable_pnpm
+      # (a) Commit identity: the provider's claimed SHA must equal the actual
+      # checkout HEAD — verification of anything else must never authorize a
+      # publish of this commit.
+      - |
+        if [ -z "$CI_COMMIT_SHA" ]; then
+          echo "[verify] FATAL: CI_COMMIT_SHA is empty — cannot certify commit identity" >&2
+          exit 1
+        fi
+        CHECKOUT_SHA="$(git rev-parse HEAD 2>/dev/null || true)"
+        if [ -z "$CHECKOUT_SHA" ]; then
+          echo "[verify] FATAL: git rev-parse HEAD returned nothing — cannot certify commit identity" >&2
+          exit 1
+        fi
+        if [ "$CI_COMMIT_SHA" != "$CHECKOUT_SHA" ]; then
+          echo "[verify] FATAL: provider commit ($CI_COMMIT_SHA) != checkout HEAD ($CHECKOUT_SHA)" >&2
+          exit 1
+        fi
+        echo "[verify] commit identity confirmed: $CHECKOUT_SHA"
+      # (b) Canonical terminal verification. Caller-provided prerequisites the
+      # runner expects (see .woodpecker/ci.yml comments): bash/rsync for the
+      # guard stages, openssl + the pinned pi binary for the test stage. git is
+      # baked into ci-base but re-asserted here so the identity check above can
+      # never silently depend on a stale baked image. DATABASE_URL is
+      # deliberately NOT set: the canonical command must hold on the PGlite
+      # path too and never sets or requires a database itself.
+      - apk add --no-cache bash rsync openssl git
+      - npm install -g @earendil-works/[email protected]
+      - pnpm verify:release
+    depends_on:
+      - install
+
   build:
     image: *node_image
     commands:
@@ -55,6 +113,7 @@ steps:
       - pnpm build
     depends_on:
       - install
+      - verify
 
   publish-npm:
     image: *node_image
@@ -114,6 +173,7 @@ steps:
         exit 1
     depends_on:
       - build
+      - verify
 
   publish-next-npm:
     image: *node_image
@@ -192,6 +252,7 @@ steps:
         echo "[publish-next] @mosaicstack/mosaic@next resolves to $RESOLVED_VERSION"
     depends_on:
       - build
+      - verify
 
   # TODO: Uncomment when ready to publish to npmjs.org
   # publish-npmjs:
@@ -205,6 +266,7 @@ steps:
   #     - bash scripts/publish-npmjs.sh
   #   depends_on:
   #     - build
+  #     - verify
   #   when:
   #     - event: [tag]
 
@@ -213,9 +275,9 @@ steps:
     when: *image_build_when
     environment:
       REGISTRY_USER:
-        from_secret: gitea_username
+        from_secret: REGISTRY_USERNAME
       REGISTRY_PASS:
-        from_secret: gitea_password
+        from_secret: REGISTRY_PASSWORD
       CI_COMMIT_BRANCH: ${CI_COMMIT_BRANCH}
       CI_COMMIT_TAG: ${CI_COMMIT_TAG}
       CI_COMMIT_SHA: ${CI_COMMIT_SHA}
@@ -242,15 +304,16 @@ steps:
         /kaniko/executor --context . --dockerfile docker/gateway.Dockerfile $DESTINATIONS
     depends_on:
       - build
+      - verify
 
   build-appservice:
     image: gcr.io/kaniko-project/executor:debug
     when: *main_image_build_when
     environment:
       REGISTRY_USER:
-        from_secret: gitea_username
+        from_secret: REGISTRY_USERNAME
       REGISTRY_PASS:
-        from_secret: gitea_password
+        from_secret: REGISTRY_PASSWORD
       CI_COMMIT_BRANCH: ${CI_COMMIT_BRANCH}
       CI_COMMIT_TAG: ${CI_COMMIT_TAG}
       CI_COMMIT_SHA: ${CI_COMMIT_SHA}
@@ -268,15 +331,16 @@ steps:
         /kaniko/executor --context . --dockerfile docker/appservice.Dockerfile $DESTINATIONS
     depends_on:
       - build
+      - verify
 
   build-web:
     image: gcr.io/kaniko-project/executor:debug
     when: *main_image_build_when
     environment:
       REGISTRY_USER:
-        from_secret: gitea_username
+        from_secret: REGISTRY_USERNAME
       REGISTRY_PASS:
-        from_secret: gitea_password
+        from_secret: REGISTRY_PASSWORD
       CI_COMMIT_BRANCH: ${CI_COMMIT_BRANCH}
       CI_COMMIT_TAG: ${CI_COMMIT_TAG}
       CI_COMMIT_SHA: ${CI_COMMIT_SHA}
@@ -294,3 +358,4 @@ steps:
         /kaniko/executor --context . --dockerfile docker/web.Dockerfile $DESTINATIONS
     depends_on:
       - build
+      - verify

AGENTS.md

@@ -81,6 +81,73 @@ pnpm format:check  # Prettier check
 pnpm build         # Build all packages and applications
 ```
 
+## Branch Model and Merge Process — `main` and `next` (CANONICAL)
+
+**Every contribution targets `next` first. No exceptions.** Features, fixes, tests,
+docs, and policy changes all take the same route; urgency changes queue priority,
+never the route. Agents never commit to or merge into `main`.
+
+| Branch | Role                                                             | Who merges into it                                                          |
+| ------ | ---------------------------------------------------------------- | --------------------------------------------------------------------------- |
+| `next` | Integration trunk — the only PR target for contributions         | The designated merge-gate agent, after all gates pass. Never the PR author. |
+| `main` | Stable/release line — receives promotion merges from `next` only | Jason only (or an agent he explicitly delegates for a named promotion).     |
+
+### Contribution sequencing (in order, no skipping)
+
+1. **Issue first.** Work is tracked in a Gitea issue before a branch exists. The
+   issue number appears in the branch name and the PR body.
+2. **Branch from the current `origin/next` head.** Name it
+   `feat/…`, `fix/…`, `docs/…`, or `test/…` with the issue number
+   (e.g. `docs/1214-branch-process`). Record the base SHA in the PR body.
+3. **Develop with evidence.** Applicable tests accompany the change. Hooks are
+   never bypassed (`--no-verify` is prohibited). Stage explicit paths — never
+   `git add -A`.
+4. **Open the PR against `next`.** The body states: scope, base SHA,
+   verification commands with results, and any known pre-existing failures on
+   the base — documented, not retried to green and not absorbed silently.
+5. **CI must be terminal-green on the exact head.** All bounded Woodpecker
+   steps succeed (`verify-terminal-green` contract). Pipelines for fork PRs
+   start `blocked`; a maintainer approves the run — approving CI is not
+   approving the PR.
+6. **Independent review. Self-merge is prohibited** — for every agent, on every
+   PR, including trivial ones. Where the change touches protected or
+   contract-bearing content, the reviewer verifies the exact head
+   (exact-byte/exact-blob comparison), not a description of it. An `AMEND`
+   verdict returns the PR to its author; the reviewer's gate stays held until
+   a fresh exact head passes.
+7. **Merge into `next`** happens only after CI green + review pass, pinned to
+   the reviewed head SHA (a post-review push voids the review).
+8. **Promotion `next` → `main`** is a deliberate, Jason-owned reconciliation
+   merge — not part of any contribution's lifecycle. Contributors are done at
+   step 7.
+
+### Responsibilities
+
+- **Contributor** — base pinning, green CI, evidence in the PR body,
+  responding to AMEND verdicts, never merging own work.
+- **Reviewer / merge gate** — independent verification on the exact head;
+  holds and lifts gates; executes the merge into `next`.
+- **Orchestrator / adjudicator** — cross-PR sequencing, disposition when PRs
+  collide, conflict adjudication.
+- **Jason** — `next` → `main` promotions, merge-authority grants, collaborator
+  and token provisioning. Agents cannot grant themselves or each other any of
+  these.
+
+### Hotfixes and divergence
+
+- A hotfix follows the same path: branch from `next`, PR to `next`, gates,
+  merge, then an expedited Jason-owned promotion if `main` needs it urgently.
+  Committing the fix to `main` directly is prohibited even under pressure.
+- **Never land work on `main` that is not on `next`.** This has happened
+  (issue #1152's goal controller reached `main` without reaching `next`) and
+  every later PR paid for it. If it happens anyway: transplant the work onto
+  a `next`-based branch with provenance-preserving commits
+  (`git cherry-pick -x` or explicit SHA references in the messages), PR it
+  through the normal gates, and let promotion re-align `main`. Do not
+  hand-patch `main` to compensate.
+- Force-pushing a branch you do not own is prohibited; rebasing your own PR
+  branch is fine before review, and voids any review already given.
+
 ## Database and Local Runtime Safety
 
 - Current local data-layer work uses in-process PGlite; leave `DATABASE_URL` unset.

README.md

@@ -146,9 +146,9 @@ mosaic brain tasks
 mosaic brain conversations
 
 # Agent forge pipeline
-mosaic forge run
+mosaic forge run [--simulate]        # fails closed (FORGE_NO_EXECUTOR) with no executor wired; --simulate for typed simulated runs
 mosaic forge status
-mosaic forge resume
+mosaic forge resume [--simulate]     # same fail-closed rule as forge run
 mosaic forge personas
 
 # Structured logging
@@ -339,7 +339,7 @@ The framework is the bash-based standards layer installed to every developer mac
 ├── bin/mosaic             ← Unified launcher (claude, codex, opencode, pi, yolo)
 ├── guides/                ← E2E delivery, orchestrator protocol, PRD, etc.
 ├── runtime/               ← Per-runtime configs (claude/, codex/, opencode/, pi/)
-├── skills/                ← Universal skills (synced from agent-skills repo)
+├── skills/                ← Universal skills (shipped with the framework package)
 ├── tools/                 ← Tool suites (orchestrator, git, quality, prdy, etc.)
 └── memory/                ← Persistent agent memory (preserved across upgrades)
 ```

REPORT-T1.md

@@ -1,94 +0,0 @@
-# T1 report: canonical ungated Claude base and lease overlay
-
-## Changed
-
-- Replaced `packages/mosaic/framework/runtime/claude/settings.json` with the canonical ungated base. It retains the model, QA hooks, plugins, command allowlist, permissions, and `mcpServers.sequential-thinking`.
-- Added `packages/mosaic/framework/runtime/claude/lease-overlay.json`. It contains only `hooks` and the six removed lease hook entries.
-- Added the byte-identical pre-split source fixture at `packages/mosaic/src/runtime/fixtures/claude-settings.gated.pre-split.json`.
-- Added `packages/mosaic/src/runtime/claude-settings-base.spec.ts`.
-
-`framework-manifest.txt` already declares `runtime/**`, so the new overlay is framework-owned and shipped without a manifest change.
-
-## Lease-hook enumeration
-
-The actual template has six lease hook entries, matching fred's refined boundary:
-
-1. `PreToolUse` matcher `.*`: `mutator-gate.py`
-2. `Stop`: one combined command containing `receipt-observer-client.py` then `promote-complete.py`
-3. `UserPromptSubmit` matcher `^/mosaic-promote$`: `promote-begin.py`
-4. `PreCompact`: `revoke-lease.py --reason pre-compact`
-5. `SessionStart` matcher `compact`: `revoke-lease.py --reason session-start-compact`
-6. `SessionStart` matcher `resume|clear`: `revoke-lease.py --reason session-start-rollover --bump-generation`
-
-There is no delta from the refined six-entry enumeration. The Stop entry contains the receipt-observer and promote-complete commands together, rather than as two separate hook objects.
-
-## Tests and checks
-
-`pnpm install --frozen-lockfile` was run first because `node_modules` was absent. It completed successfully.
-
-Red-first run before artifacts existed:
-
-```text
-RUN  v2.1.9 .../packages/mosaic
-❯ src/runtime/claude-settings-base.spec.ts (4 tests | 4 failed)
-× keeps every lease command out of the ungated base
-  → mutator-gate: expected true to be false
-× reconstructs the pre-split gated hooks while retaining the canonical MCP correction
-  → ENOENT: .../lease-overlay.json
-× ships sequential-thinking in the base
-  → expected undefined to deeply equal { 'sequential-thinking': ... }
-× limits the overlay to lease hook entries
-  → ENOENT: .../lease-overlay.json
-```
-
-Final focused acceptance run:
-
-```text
-RUN  v2.1.9 .../packages/mosaic
-✓ src/runtime/claude-settings-base.spec.ts (4 tests) 19ms
-Test Files  1 passed (1)
-Tests  4 passed (4)
-```
-
-`pnpm --filter @mosaicstack/mosaic lint` passed:
-
-```text
-> @mosaicstack/[email protected] lint
-> eslint src
-```
-
-`pnpm --filter @mosaicstack/mosaic typecheck` failed on pre-existing workspace resolution and unrelated package errors. The new spec no longer appears in the error list. Initial failures include missing `@mosaicstack/{brain,forge,log,macp,memory,queue,storage,quality-rails,db,config,prdy,types}` declarations, followed by existing `fleet-backlog.ts`, `gateway-doctor.ts`, and TUI implicit-`any` errors. Exit status: 2.
-
-A focused legacy consumer run confirms an existing assumption that `settings.json` itself is gated:
-
-```text
-pnpm --filter @mosaicstack/mosaic exec vitest run src/mutator-gate/mutator-gate.acceptance.spec.ts
-❯ src/mutator-gate/mutator-gate.acceptance.spec.ts (20 tests | 6 failed)
-× non-dangerous parser residual is denied by the global all-tools hook without a lease
-  → expected all-tools mutator-gate command in settings.json
-× Claude and Pi compaction observer wiring is complete and fail-closed
-  → expected PreCompact/SessionStart revoke-lease hooks in settings.json
-```
-
-The other four failures in that focused run reported `STALE_GENERATION` where the test expected `MUTATOR_UNVERIFIED`, plus one successful-gate assertion. I did not redesign this legacy suite because the task explicitly says to report consumers that assume the base is gated.
-
-## Consumers found
-
-Direct `runtime/claude/settings.json` path consumers found by the required repository grep:
-
-- `packages/mosaic/framework/tools/_scripts/mosaic-link-runtime-assets`: copies the base to `~/.claude/settings.json`.
-- `packages/mosaic/src/commands/install-ordering-guard.ts` and `.spec.ts`: documentation and behavior assume the source embeds enforcement hooks.
-- `packages/mosaic/framework/tools/_scripts/test-install-ordering-guard.sh`: comments and assertions expect `mutator-gate.py` and `receipt-observer-client.py` in the base.
-- `packages/mosaic/src/mutator-gate/mutator-gate.acceptance.spec.ts`: reads the base and asserts mutator, promotion, and compaction lease wiring.
-- `packages/mosaic/src/lease-broker/promotion_trigger_unittest.py`: reads the base and asserts promotion wiring.
-- `packages/mosaic/src/lease-broker/recovery_runtime_unittest.py`: reads the base.
-- `packages/mosaic/src/runtime/update-checker.ts` and `.spec.ts`: references the path in settings wiring/update checks.
-- Documentation-only references: `docs/compaction-refresh/probes/p6_constrained_recovery.py`, `docs/plans/agent-reflection-loop-PRD.md`, `docs/tasks/544-agent-reflection-loop.md`, and the framework QA documentation/scripts found by grep.
-
-I did not change these consumers. The install/link and lease acceptance consumers must be taught to select and compose `lease-overlay.json` when a gated promotion seat is requested. That composition behavior is outside T1.
-
-## Ambiguity handled
-
-The exact pre-split template fixture has no `mcpServers` key (SHA-256 `44e74ea1e9d424fffa020ee666402662ac856b88bf6ae7f3b8931eed29dc75a4`). The task simultaneously requires a byte-for-byte pre-split fixture, `mcpServers.sequential-thinking` in the base, and `deep-merge(base, overlay) == original`. Those three conditions cannot all hold because a merge cannot remove the required MCP key.
-
-The acceptance test preserves the exact fixture and asserts that the normalized merge equals the pre-split template plus the required canonical `mcpServers.sequential-thinking` correction. It verifies all original hook content is reconstructed and the base carries the required MCP. Production three-layer merge semantics remain W-F1 work.

REPORT-T2.md

@@ -1,102 +0,0 @@
-# REPORT-T2
-
-Date: 2026-08-13 11:29 CDT
-Branch: `feat/wf-fleet-t2-launch`
-Base: `216cd722`
-Issue: #1209
-
-## What changed
-
-- Added `mosaic fleet launch <name> [--dry-run]` in `packages/mosaic/src/commands/fleet-launch-command.ts` and registered it on the existing fleet command.
-- Added strict schema-one parsing for the user-owned `~/.mosaic/fleet/agents/<name>/profile.json`:
-  - required `schema` and `harness`
-  - default bundle `primary`
-  - optional `model`, `overlay`, `plugins`, `skills`, and string-valued `env`
-  - unknown-key refusal naming the key
-  - dedicated `SCHEMA_TOO_NEW` code and upgrade guidance
-- Added the three-layer settings composer. Objects merge recursively, scalars use the higher layer, arrays replace, and `null` deletes a key. The selected agent overlay defaults to no overlay when the profile field is absent.
-- Writes canonical merged settings to `<agent-home>/settings.json` and the future harvest comparison snapshot to `<agent-dir>/settings.generated.json`.
-- Resolves `primary` to its named bundle, reads an optional account email, and reports forms such as `primary -> fred_example.com ([email protected])`.
-- Validates credential targets with `lstat`, rejects symlink credential files, resolves and checks containment under the harness auth root, and refuses a real credential file at the seat-link path as first-auth state.
-- Installs selected plugin and skill entries as seat-local symlinks, prunes stale symlinks, and refuses real objects instead of deleting them.
-- Builds a declared seat environment with the harness home variable, `MOSAIC_AGENT_NAME`, and profile environment entries. Mechanical values override conflicting profile entries.
-- Extended `launch.ts` so `harnessHome()` accepts fleet context and remains the home-resolution seam. The fleet launcher uses the existing runtime preflight, prompt, ledger, lease-gated, and process execution path over a minimal ambient environment.
-- Added deterministic dry-run output containing source layers, merged settings, output and snapshot paths, resolved bundle, symlink plans, declared environment, and harness argv.
-- Added 17 focused tests, including the required merge, schema, A3, dry-run snapshot, managed-link, command dry-run, execution-seam, and non-zero failure cases.
-
-## Reconciliation decisions and contradictions
-
-### Prominent contradiction: roster registries do not contain the frozen launch schema
-
-The existing code has two other profile/registry concepts:
-
-- `fleet-profiles.ts` models system-type YAML roster templates. Its `FleetProfile` has no harness bundle, overlay, plugin, skill, or seat environment fields.
-- roster-v2 models topology and lifecycle. It requires class, provider, reasoning, tool policy, working directory, lifecycle, and launch-yolo fields that schema-one `profile.json` does not contain.
-
-Deriving a complete roster-v2 member from the frozen per-agent profile is therefore not possible without inventing values. Launch now reads only the per-agent `profile.json` and does not require roster-v2 or the legacy v1 roster. roster-v2 remains the existing lifecycle/topology registry. No second launch registry was introduced.
-
-The pre-existing `resolveFleetIdentity()` path requires a legacy roster and a secure tmux helper whenever `MOSAIC_AGENT_NAME` is present during contract composition. For profile-backed launch, `launch.ts` excludes roster identity keys only from the contract-build environment, then exports the declared profile seat identity to the harness process. Legacy root runtime launches retain the existing roster-backed behavior. This is the smallest reconciliation that allows profile-only launch without fabricating roster-v2 fields.
-
-### Historical whole-store plugin link
-
-The prototype used a whole `plugins` directory symlink, while this task requires selected entry links and pruning. Launch refuses that historical shape with an explicit migration message. It does not delete or silently convert the whole-store link.
-
-### Existing `FleetProfile` name
-
-The system-type YAML `FleetProfile` remains unchanged. The new type is named `FleetAgentLaunchProfile` to keep the concepts separate while treating per-agent `profile.json` as the launch SSOT.
-
-## Ambiguities and bounded choices
-
-- The design does not freeze the generated snapshot filename. This implementation uses `settings.generated.json` in the agent directory, beside the hidden harness home.
-- The design explicitly identifies Claude `.credentials.json` and Pi `auth.json`. Codex and OpenCode use `auth.json` in the filename map, matching their harness-home composition shape, but no real credential launch was performed in this task.
-- Full interactive harvest-back disposition is not implemented. The task asks to store the generated snapshot for the future diff, and this change does that.
-- A machine descriptor file and content digests were not added. Dry-run and execution consume one resolved in-memory composition, and dry-run prints that composition.
-- No real harness process or real operator home was used. Every new filesystem test uses a temporary fixture root.
-
-## Test run
-
-Dependency install and build:
-
-```text
-$ pnpm install --frozen-lockfile
-Scope: all 28 workspace projects
-Lockfile is up to date, resolution step is skipped
-Done in 4.7s using pnpm v10.6.2
-
-$ pnpm --filter @mosaicstack/mosaic... build
-Scope: 13 of 28 workspace projects
-packages/mosaic build: Done
-```
-
-Focused and touched integration tests:
-
-```text
-$ pnpm --filter @mosaicstack/mosaic exec vitest run src/commands/fleet-launch-command.spec.ts src/commands/launch.spec.ts src/commands/fleet.spec.ts
-Test Files  3 passed (3)
-Tests       256 passed (256)
-```
-
-Typecheck and lint:
-
-```text
-$ pnpm --filter @mosaicstack/mosaic typecheck
-> tsc --noEmit
-(exit 0)
-
-$ pnpm exec eslint packages/mosaic/src/commands/fleet-launch-command.ts packages/mosaic/src/commands/fleet-launch-command.spec.ts packages/mosaic/src/commands/launch.ts packages/mosaic/src/commands/fleet.ts packages/mosaic/src/commands/fleet.spec.ts
-(exit 0)
-
-$ pnpm exec prettier --check packages/mosaic/src/commands/fleet-launch-command.ts packages/mosaic/src/commands/fleet-launch-command.spec.ts packages/mosaic/src/commands/launch.ts packages/mosaic/src/commands/fleet.ts packages/mosaic/src/commands/fleet.spec.ts
-Checking formatting...
-All matched files use Prettier code style!
-```
-
-Package-wide Vitest result:
-
-```text
-$ pnpm --filter @mosaicstack/mosaic exec vitest run
-Test Files  1 failed | 83 passed (84)
-Tests       4 failed | 1535 passed (1539)
-```
-
-All four failures are in `src/mutator-gate/mutator-gate.acceptance.spec.ts`. Three expected `MUTATOR_UNVERIFIED` but received `STALE_GENERATION`; one runtime-gate assertion expected status zero and received status two. An isolated rerun produced the same four failures. I did not confirm whether they predate this branch. The focused launch, fleet, and typecheck runs are green.

REPORT-T3.md

@@ -1,46 +0,0 @@
-# T3 report: `mosaic fleet agent new`
-
-## Changed
-
-- Added `packages/mosaic/src/fleet/fleet-agent-scaffold.ts`.
-  - Creates user-owned seats at `~/.mosaic/fleet/agents/<name>` (test seam: `fleetDataHome`, environment default: `MOSAIC_DATA_HOME`).
-  - Writes schema-one `profile.json` with default `harness: "claude"`, `bundle: "primary"`, optional `model`, `overlay: "overlay.json"`, and mandatory `env.MOSAIC_AGENT_NAME`.
-  - Writes a positive `SOUL.md` identity and materializes that identity in `.claude/CLAUDE.md` or `.pi/AGENTS.md`.
-  - Writes `overlay.json` as `{}`. Claude homes get `.claude.json` with `hasCompletedOnboarding: true` and `theme: "dark"`. No settings file is composed.
-  - Creates the appropriate credential symlink (`.credentials.json` for Claude, `auth.json` for Pi), allowing an intentional dangling destination and reporting it at the command surface.
-  - Compares every existing object (including link targets as link text), succeeds only byte-identically, and otherwise refuses with the differing paths.
-- Added `packages/mosaic/src/commands/fleet-agent-scaffold-command.ts` and wired `fleet agent new <name> [--harness claude|pi] [--bundle B] [--model M]` in `packages/mosaic/src/commands/fleet.ts`.
-- Added `packages/mosaic/src/commands/fleet-agent-scaffold-command.spec.ts` with temp-root-only coverage: exact Claude/Pi layouts, literal quote/backtick/`$( )` handling, unsafe names and option failures, idempotence, changed-file refusal, and credential-link comparison.
-
-## Reconciliation
-
-`fleet-agent-crud-command.ts` currently registers roster-v2 `get/create/update/delete/plan` directly under `mosaic fleet`; it has no `agent new` command or profile schema. T3 adds an `agent` namespace for the profile-owned user-data scaffold and leaves roster-v2 CRUD unchanged.
-
-No roster projection is created. Current roster-v2 requires fields that cannot be derived from the new profile (`class`, provider, working directory, reasoning, tool policy, lifecycle), while no current `mosaic fleet launch <name>` consumes these profiles. Writing such a roster entry would create the forbidden second registry and invent semantics. The profile is therefore the sole state created here. When the launcher owns profile-to-roster projection, it must derive it there and emit the required actionable unscaffolded-name message.
-
-## Validation
-
-```text
-$ pnpm install --frozen-lockfile
-Done in 4.1s using pnpm v10.6.2
-
-$ pnpm --filter @mosaicstack/mosaic exec vitest run src/commands/fleet-agent-scaffold-command.spec.ts
-✓ src/commands/fleet-agent-scaffold-command.spec.ts (13 tests) 28ms
-Test Files  1 passed (1)
-Tests       13 passed (13)
-
-$ pnpm --filter @mosaicstack/mosaic exec eslint src/fleet/fleet-agent-scaffold.ts src/commands/fleet-agent-scaffold-command.ts src/commands/fleet-agent-scaffold-command.spec.ts src/commands/fleet.ts
-(exit 0)
-
-$ pnpm exec prettier --check packages/mosaic/src/fleet/fleet-agent-scaffold.ts packages/mosaic/src/commands/fleet-agent-scaffold-command.ts packages/mosaic/src/commands/fleet-agent-scaffold-command.spec.ts packages/mosaic/src/commands/fleet.ts
-All matched files use Prettier code style!
-
-$ git diff --check
-(exit 0)
-```
-
-`pnpm --filter @mosaicstack/mosaic typecheck` remains blocked by pre-existing unresolved workspace package entries (`@mosaicstack/brain`, `@mosaicstack/db`, `@mosaicstack/types`, and others). The typecheck output had no diagnostics naming T3 files. Running the pre-existing CRUD command spec is blocked by the same `@mosaicstack/db` Vite resolution failure through `fleet-backlog.ts`.
-
-## Skipped ambiguity
-
-The design asks for a generated harness-home `settings.json` as part of an earlier generic home-template description, but the task explicitly says composed settings are left to launch. T3 creates no `settings.json`; launch composition remains the owner.

apps/gateway/src/__tests__/cross-user-isolation.test.ts

@@ -190,7 +190,13 @@ beforeEach((ctx) => {
 });
 
 afterAll(async () => {
-  if (!handle) return;
+  // Cleanup only when the fixture actually installed rows. `handle` is set
+  // before the first query (createDb connects lazily), so on an unreachable
+  // database `handle` is truthy while nothing was inserted — cleanup must
+  // honor `dbAvailable` or the skip path fails the file with ECONNREFUSED in
+  // afterAll (caught live by the publish pipeline's no-DATABASE_URL verify
+  // step, pipeline 2486).
+  if (!handle || !dbAvailable) return;
   const db = handle.db;
 
   // Delete in dependency order (FK constraints)

apps/gateway/src/__tests__/required-security-wiring.test.ts

@@ -0,0 +1,332 @@
+import { type Type } from '@nestjs/common';
+import { Test, type TestingModule } from '@nestjs/testing';
+import type { SlashCommandPayload } from '@mosaicstack/types';
+import { describe, expect, it, vi } from 'vitest';
+import { AgentService, type AgentSession } from '../agent/agent.service.js';
+import { ProviderService } from '../agent/provider.service.js';
+import { AppModule } from '../app.module.js';
+import { CommandAuthorizationService } from '../commands/command-authorization.service.js';
+import { CommandExecutorService } from '../commands/command-executor.service.js';
+import { CommandsModule } from '../commands/commands.module.js';
+import { CommandRuntimeApprovalVerifier } from '../commands/runtime-approval-verifier.js';
+import { PreferencesModule } from '../preferences/preferences.module.js';
+import { SystemOverrideService } from '../preferences/system-override.service.js';
+
+const fakeDb = {
+  $client: { exec: async (): Promise<void> => {} },
+  execute: async (): Promise<{ rows: unknown[] }> => ({ rows: [] }),
+  select: () => ({
+    from: () => ({
+      where: async (): Promise<Array<{ count: number }>> => [{ count: 1 }],
+    }),
+  }),
+  insert: () => ({ values: async (): Promise<void> => {} }),
+};
+
+const fakeProviderService = {
+  onModuleInit: async (): Promise<void> => {},
+  onModuleDestroy: (): void => {},
+  getRegistry: () => ({ getAvailable: () => [], getAll: () => [], find: () => undefined }),
+  getDefaultModel: () => undefined,
+  listAvailableModels: () => [],
+  listProviders: () => [],
+  getAdapter: () => undefined,
+  getProvidersHealth: () => [],
+};
+
+function compileRealAppGraph(): Promise<TestingModule> {
+  return Test.createTestingModule({ imports: [AppModule] })
+    .overrideProvider('DB_HANDLE')
+    .useValue({ db: fakeDb, close: async (): Promise<void> => {} })
+    .overrideProvider('DB')
+    .useValue(fakeDb)
+    .overrideProvider('STORAGE_ADAPTER')
+    .useValue({
+      name: 'required-security-wiring-test',
+      migrate: async (): Promise<void> => {},
+      close: async (): Promise<void> => {},
+    })
+    .overrideProvider('AUTH')
+    .useValue({})
+    .overrideProvider('BRAIN')
+    .useValue({ conversations: {}, agents: {} })
+    .overrideProvider('LOG_SERVICE')
+    .useValue({})
+    .overrideProvider('MEMORY')
+    .useValue({})
+    .overrideProvider('MEMORY_ADAPTER')
+    .useValue({})
+    .overrideProvider(ProviderService)
+    .useValue(fakeProviderService)
+    .compile();
+}
+
+function providerToken(provider: unknown): unknown {
+  return typeof provider === 'function' ? provider : (provider as { provide?: unknown })?.provide;
+}
+
+interface MaskingConsumer {
+  moduleType: Type<unknown>;
+  token: Type<unknown>;
+  useValue: object;
+}
+
+async function compileWithoutProvider(
+  moduleType: Type<unknown>,
+  missingToken: Type<unknown>,
+  maskingConsumer: MaskingConsumer,
+): Promise<{ error: unknown; moduleRef: TestingModule | undefined }> {
+  const touchedModules = new Set([moduleType, maskingConsumer.moduleType]);
+  const originals = Array.from(touchedModules, (touchedModule: Type<unknown>) => ({
+    moduleType: touchedModule,
+    providers: (Reflect.getMetadata('providers', touchedModule) ?? []) as unknown[],
+    exports: (Reflect.getMetadata('exports', touchedModule) ?? []) as unknown[],
+  }));
+
+  for (const original of originals) {
+    const providers = original.providers.flatMap((provider: unknown): unknown[] => {
+      const token = providerToken(provider);
+      if (original.moduleType === moduleType && token === missingToken) return [];
+      if (original.moduleType === maskingConsumer.moduleType && token === maskingConsumer.token) {
+        return [{ provide: maskingConsumer.token, useValue: maskingConsumer.useValue }];
+      }
+      return [provider];
+    });
+    const exports = original.exports.filter(
+      (exported: unknown): boolean =>
+        original.moduleType !== moduleType || providerToken(exported) !== missingToken,
+    );
+    Reflect.defineMetadata('providers', providers, original.moduleType);
+    Reflect.defineMetadata('exports', exports, original.moduleType);
+  }
+
+  let moduleRef: TestingModule | undefined;
+  let error: unknown;
+  try {
+    moduleRef = await compileRealAppGraph();
+  } catch (caught: unknown) {
+    error = caught;
+  } finally {
+    for (const original of originals) {
+      Reflect.defineMetadata('providers', original.providers, original.moduleType);
+      Reflect.defineMetadata('exports', original.exports, original.moduleType);
+    }
+  }
+  return { error, moduleRef };
+}
+
+async function closeIfCompiled(moduleRef: TestingModule | undefined): Promise<void> {
+  if (moduleRef) await moduleRef.close();
+}
+
+describe('required security wiring — real AppModule startup refusal', () => {
+  it('FL-01 positive control: the real graph compiles when CommandAuthorizationService is bound', async () => {
+    const moduleRef = await compileRealAppGraph();
+    try {
+      expect(moduleRef.get(CommandAuthorizationService, { strict: false })).toBeInstanceOf(
+        CommandAuthorizationService,
+      );
+    } finally {
+      await moduleRef.close();
+    }
+  });
+
+  it('FL-01 negative control: absence read as permission is refused at module compilation', async () => {
+    const { error, moduleRef } = await compileWithoutProvider(
+      CommandsModule,
+      CommandAuthorizationService,
+      {
+        moduleType: CommandsModule,
+        token: CommandRuntimeApprovalVerifier,
+        useValue: {},
+      },
+    );
+    await closeIfCompiled(moduleRef);
+
+    expect(
+      error,
+      'absence read as permission: AppModule compilation accepted a missing CommandAuthorizationService binding',
+    ).toBeInstanceOf(Error);
+    if (!(error instanceof Error)) return;
+    expect(error.message).toContain('CommandExecutorService');
+    expect(error.message).toContain('CommandAuthorizationService');
+  });
+
+  it('FL-11 positive control: the real graph compiles when SystemOverrideService is bound', async () => {
+    const moduleRef = await compileRealAppGraph();
+    try {
+      expect(moduleRef.get(SystemOverrideService, { strict: false })).toBeInstanceOf(
+        SystemOverrideService,
+      );
+    } finally {
+      await moduleRef.close();
+    }
+  });
+
+  it('FL-11 negative control: absence read as permission is refused at module compilation', async () => {
+    const { error, moduleRef } = await compileWithoutProvider(
+      PreferencesModule,
+      SystemOverrideService,
+      {
+        moduleType: CommandsModule,
+        token: CommandExecutorService,
+        useValue: {},
+      },
+    );
+    await closeIfCompiled(moduleRef);
+
+    expect(
+      error,
+      'absence read as permission: AppModule compilation accepted a missing SystemOverrideService binding',
+    ).toBeInstanceOf(Error);
+    if (!(error instanceof Error)) return;
+    expect(error.message).toContain('AgentService');
+    expect(error.message).toContain('SystemOverrideService');
+  });
+});
+
+const actorScope = { userId: 'security-user', tenantId: 'security-tenant' };
+const conversationId = 'security-conversation';
+
+function directExecutorWithoutAuthorization(systemOverrideSet: ReturnType<typeof vi.fn>) {
+  const registry = {
+    getManifest: vi.fn(() => ({
+      version: 1,
+      commands: [
+        {
+          name: 'system',
+          aliases: [],
+          description: 'Set instruction authority',
+          scope: 'agent' as const,
+          execution: 'socket' as const,
+          available: true,
+        },
+      ],
+      skills: [],
+    })),
+  };
+  return new CommandExecutorService(
+    registry as never,
+    { getSession: vi.fn() } as never,
+    { set: systemOverrideSet, clear: vi.fn() } as never,
+    { collect: vi.fn() } as never,
+    null,
+    { agents: {} } as never,
+    null,
+    null,
+    { getServerStatuses: vi.fn(() => []), getToolDefinitions: vi.fn(() => []) } as never,
+    undefined as never,
+  );
+}
+
+function directAgentWithoutSystemOverride(piPrompt: ReturnType<typeof vi.fn>): {
+  service: AgentService;
+  session: AgentSession;
+} {
+  const service = new AgentService(
+    {
+      getDefaultModel: vi.fn(() => null),
+      getRegistry: vi.fn(() => ({})),
+      findModel: vi.fn(),
+      listAvailableModels: vi.fn(() => []),
+    } as never,
+    {} as never,
+    {} as never,
+    { available: false } as never,
+    {} as never,
+    { getToolDefinitions: vi.fn(() => []) } as never,
+    { loadForSession: vi.fn(async () => ({ metaTools: [], promptAdditions: [] })) } as never,
+    undefined as never,
+    null,
+    { collect: vi.fn().mockResolvedValue(undefined) } as never,
+    null,
+  );
+  const session = {
+    id: conversationId,
+    provider: 'test-provider',
+    modelId: 'test-model',
+    piSession: { prompt: piPrompt },
+    listeners: new Set(),
+    unsubscribe: vi.fn(),
+    createdAt: Date.now(),
+    promptCount: 0,
+    channels: new Set(),
+    skillPromptAdditions: [],
+    sandboxDir: process.cwd(),
+    allowedTools: null,
+    userId: actorScope.userId,
+    tenantId: actorScope.tenantId,
+    metrics: {
+      tokens: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
+      modelSwitches: 0,
+      messageCount: 0,
+      lastActivityAt: new Date(0).toISOString(),
+    },
+  } as unknown as AgentSession;
+  const internals = service as unknown as { sessions: Map<string, AgentSession> };
+  internals.sessions.set(conversationId, session);
+  return { service, session };
+}
+
+describe('required security wiring — malformed direct absence has zero effects', () => {
+  it('FL-01 refuses command execution before any command effect when authorization is absent', async () => {
+    const systemOverrideSet = vi.fn().mockResolvedValue(undefined);
+    const executor = directExecutorWithoutAuthorization(systemOverrideSet);
+    const payload: SlashCommandPayload = {
+      command: 'system',
+      args: 'authority that must not be stored',
+      conversationId,
+    };
+    let error: unknown;
+
+    try {
+      await executor.execute(payload, actorScope);
+    } catch (caught: unknown) {
+      error = caught;
+    }
+
+    expect
+      .soft(
+        error,
+        'absence read as permission: direct executor accepted missing command authorization',
+      )
+      .toBeInstanceOf(Error);
+    expect
+      .soft(
+        systemOverrideSet,
+        'absence read as permission: command effect occurred without command authorization',
+      )
+      .not.toHaveBeenCalled();
+  });
+
+  it('FL-11 refuses prompt execution before any provider or session effect when system override authority is absent', async () => {
+    const piPrompt = vi.fn().mockResolvedValue(undefined);
+    const { service, session } = directAgentWithoutSystemOverride(piPrompt);
+    let error: unknown;
+
+    try {
+      await service.prompt(conversationId, 'must not reach provider', actorScope);
+    } catch (caught: unknown) {
+      error = caught;
+    }
+
+    expect
+      .soft(
+        error,
+        'absence read as permission: direct session accepted missing system override authority',
+      )
+      .toBeInstanceOf(Error);
+    expect
+      .soft(
+        piPrompt,
+        'absence read as permission: provider prompt occurred without system override authority',
+      )
+      .not.toHaveBeenCalled();
+    expect
+      .soft(
+        session.promptCount,
+        'absence read as permission: session state changed without system override authority',
+      )
+      .toBe(0);
+  });
+});

apps/gateway/src/agent/__tests__/agent-service-ownership.test.ts

@@ -26,7 +26,7 @@ function makeService(operatorMemory: unknown = null): AgentService {
     {} as never,
     { getToolDefinitions: vi.fn(() => []) } as never,
     { loadForSession: vi.fn(async () => ({ metaTools: [], promptAdditions: [] })) } as never,
-    null,
+    { get: vi.fn().mockResolvedValue(null), renew: vi.fn().mockResolvedValue(undefined) } as never,
     null,
     { collect: vi.fn().mockResolvedValue(undefined) } as never,
     operatorMemory as never,

apps/gateway/src/agent/agent.service.ts

@@ -132,9 +132,8 @@ export class AgentService implements OnModuleDestroy {
     @Inject(CoordService) private readonly coordService: CoordService,
     @Inject(McpClientService) private readonly mcpClientService: McpClientService,
     @Inject(SkillLoaderService) private readonly skillLoaderService: SkillLoaderService,
-    @Optional()
     @Inject(SystemOverrideService)
-    private readonly systemOverride: SystemOverrideService | null,
+    private readonly systemOverride: SystemOverrideService,
     @Optional()
     @Inject(PreferencesService)
     private readonly preferencesService: PreferencesService | null,
@@ -709,23 +708,22 @@ export class AgentService implements OnModuleDestroy {
       throw new Error(`No agent session found: ${sessionId}`);
     }
     this.assertSessionScope(session, scope);
-    session.promptCount += 1;
 
     // Channel attachments are untrusted URI references. Preserve exact,
     // authenticated metadata for the agent without treating it as authority.
     const attachmentContext = this.attachmentContext(attachments);
 
-    // Prepend session-scoped system override if present (renew TTL on each turn)
+    // Prepend session-scoped system override if present (renew TTL on each turn).
+    // Required instruction-authority wiring is consulted before session/provider effects.
     let effectiveMessage = `${message}${attachmentContext}`;
-    if (this.systemOverride) {
-      const override = await this.systemOverride.get(sessionId, scope);
-      if (override) {
-        effectiveMessage = `[System Override]\n${override}\n\n${effectiveMessage}`;
-        await this.systemOverride.renew(sessionId, scope);
-        this.logger.debug(`Applied system override for session ${sessionId}`);
-      }
+    const override = await this.systemOverride.get(sessionId, scope);
+    if (override) {
+      effectiveMessage = `[System Override]\n${override}\n\n${effectiveMessage}`;
+      await this.systemOverride.renew(sessionId, scope);
+      this.logger.debug(`Applied system override for session ${sessionId}`);
     }
 
+    session.promptCount += 1;
     try {
       await session.piSession.prompt(effectiveMessage);
     } catch (err) {

apps/gateway/src/commands/command-executor-p8012.spec.ts

@@ -80,6 +80,10 @@ const mockMcpClient = {
   getToolDefinitions: vi.fn(() => []),
 };
 
+const allowAuthorization = {
+  authorize: vi.fn().mockResolvedValue({ allowed: true }),
+};
+
 function buildService(
   redis: typeof mockRedis | null = mockRedis,
   mcpClient: {
@@ -98,6 +102,7 @@ function buildService(
     null,
     mockChatGateway as never,
     mcpClient as never,
+    allowAuthorization as never,
   );
 }
 

apps/gateway/src/commands/command-executor.service.ts

@@ -35,9 +35,8 @@ export class CommandExecutorService {
     @Inject(forwardRef(() => ChatGateway))
     private readonly chatGateway: ChatGateway | null,
     @Inject(McpClientService) private readonly mcpClient: McpClientService,
-    @Optional()
     @Inject(CommandAuthorizationService)
-    private readonly authorization: CommandAuthorizationService | null = null,
+    private readonly authorization: CommandAuthorizationService,
   ) {}
 
   async execute(
@@ -57,13 +56,13 @@ export class CommandExecutorService {
       };
     }
 
-    const authorization = await this.authorization?.authorize(
+    const authorization = await this.authorization.authorize(
       def,
       payload,
       userId,
       payload.approvalId,
     );
-    if (authorization && !authorization.allowed) {
+    if (!authorization.allowed) {
       return { command, conversationId, success: false, message: authorization.reason };
     }
 
@@ -171,7 +170,7 @@ export class CommandExecutorService {
     const def = this.registry
       .getManifest()
       .commands.find((command) => command.name === payload.command);
-    if (!def || !this.authorization) return null;
+    if (!def) return null;
     return this.authorization.createApproval(def, payload, scope.userId);
   }
 

apps/gateway/src/commands/commands.integration.spec.ts

@@ -55,6 +55,10 @@ const mockMcpClient = {
   reconnectServer: vi.fn().mockResolvedValue(undefined),
 };
 
+const allowAuthorization = {
+  authorize: vi.fn().mockResolvedValue({ allowed: true }),
+};
+
 // ─── Helpers ─────────────────────────────────────────────────────────────────
 
 function buildRegistry(): CommandRegistryService {
@@ -74,6 +78,7 @@ function buildExecutor(registry: CommandRegistryService): CommandExecutorService
     null, // reloadService (optional)
     null, // chatGateway (optional)
     mockMcpClient as never,
+    allowAuthorization as never,
   );
 }
 

apps/gateway/src/reload/reload.service.spec.ts

@@ -159,6 +159,7 @@ describe('ReloadService — /reload command sanitizes plugin errors', () => {
       reloadService,
       mockChatGateway as never,
       mockMcpClient as never,
+      { authorize: vi.fn().mockResolvedValue({ allowed: true }) } as never,
     );
 
     const payload: SlashCommandPayload = { command: 'reload', conversationId: 'conv-1' };

apps/web/src/components/freshness/freshness-notices.tsx

@@ -0,0 +1,110 @@
+'use client';
+
+import type { ReactElement } from 'react';
+import { formatAge, type FreshnessLabel } from '@/lib/freshness/model';
+
+/**
+ * Rendering rules for non-current freshness states (RI-5-001).
+ *
+ * - `unavailable` renders an explicit failure panel — never an empty
+ *   healthy collection.
+ * - `stale` may render last-known data, but only under a visible label
+ *   carrying source identity, snapshot version, and age.
+ * - `partial` renders the verified parts plus an explicit list of what is
+ *   missing.
+ */
+
+interface RetryableNoticeProps {
+  readonly onRetry?: () => void;
+  readonly retryLabel?: string;
+}
+
+function RetryButton({ onRetry, retryLabel }: RetryableNoticeProps): ReactElement | null {
+  if (!onRetry) return null;
+  return (
+    <button
+      type="button"
+      onClick={onRetry}
+      className="mt-2 rounded-lg border border-surface-border px-3 py-1.5 text-xs transition-colors hover:border-gray-500"
+    >
+      {retryLabel ?? 'Retry'}
+    </button>
+  );
+}
+
+export interface UnavailableDataNoticeProps extends RetryableNoticeProps {
+  /** What is unavailable, e.g. "Tasks". */
+  readonly title: string;
+  /** Optional underlying failure detail (network message, invalidation reason). */
+  readonly detail?: string | null;
+}
+
+/** Explicit `unavailable` state. Never renders as an empty healthy collection. */
+export function UnavailableDataNotice({
+  title,
+  detail,
+  onRetry,
+  retryLabel,
+}: UnavailableDataNoticeProps): ReactElement {
+  return (
+    <div role="alert" className="rounded-lg border border-error/40 px-4 py-3 text-sm">
+      <p className="font-medium text-text-primary">{title} are unavailable</p>
+      <p className="mt-1 text-text-muted">
+        This is not an empty result — the data could not be verified from the gateway.
+        {detail ? ` ${detail}` : ''}
+      </p>
+      <RetryButton onRetry={onRetry} retryLabel={retryLabel} />
+    </div>
+  );
+}
+
+export interface StaleDataNoticeProps extends RetryableNoticeProps {
+  /** Provenance of the last-known snapshot being displayed. */
+  readonly label: FreshnessLabel;
+}
+
+/**
+ * Situational-awareness banner for `stale` data: last-known data may render,
+ * but visibly labeled with source identity, snapshot version, and age.
+ */
+export function StaleDataNotice({
+  label,
+  onRetry,
+  retryLabel,
+}: StaleDataNoticeProps): ReactElement {
+  return (
+    <div role="status" className="rounded-lg border border-warning/40 px-4 py-3 text-sm">
+      <p className="font-medium text-warning">Showing last-known data — it may be out of date</p>
+      <p className="mt-1 text-xs text-text-muted">
+        Source {label.source} · snapshot v{label.version} · fetched{' '}
+        {formatAge(label.fetchedAt, Date.now())}. Verdicts derived from this data are unknown and
+        changes are disabled until it is revalidated.
+      </p>
+      <RetryButton onRetry={onRetry} retryLabel={retryLabel ?? 'Revalidate'} />
+    </div>
+  );
+}
+
+export interface PartialDataNoticeProps extends RetryableNoticeProps {
+  /** Display names of the sections whose collections are unavailable. */
+  readonly missing: readonly string[];
+}
+
+/** `partial` surface banner: verified parts render, missing parts are explicit. */
+export function PartialDataNotice({
+  missing,
+  onRetry,
+  retryLabel,
+}: PartialDataNoticeProps): ReactElement {
+  return (
+    <div role="status" className="rounded-lg border border-warning/40 px-4 py-3 text-sm">
+      <p className="font-medium text-warning">Some data could not be loaded</p>
+      <p className="mt-1 text-xs text-text-muted">
+        {missing.join(', ')} {missing.length === 1 ? 'is' : 'are'} unavailable — sections below show
+        an explicit unavailable state instead of an empty list. Derived verdicts remain unknown
+        until every collection is revalidated.
+      </p>
+      <RetryButton onRetry={onRetry} retryLabel={retryLabel ?? 'Revalidate'} />
+    </div>
+  );
+}

apps/web/src/lib/freshness/model.spec.ts

@@ -0,0 +1,324 @@
+import { describe, expect, it } from 'vitest';
+import type { Task } from '@/lib/types';
+import {
+  acceptSnapshot,
+  assertMutable,
+  canMutate,
+  combineFreshness,
+  computeDigest,
+  computeFreshness,
+  DEFAULT_FRESHNESS_POLICY,
+  formatAge,
+  type FreshSnapshot,
+  invalidationReasonLabels,
+  StaleMutationError,
+  UNKNOWN_VERDICT,
+  verdictValue,
+} from './model';
+import { validateProjectCollection, validateTaskCollection } from './validators';
+
+const NOW = 1_800_000_000_000;
+
+const policy = { ...DEFAULT_FRESHNESS_POLICY, staleAfterMs: 60_000 };
+
+const taskPayload: Task[] = [
+  {
+    id: 'task-1',
+    title: 'T1',
+    description: null,
+    status: 'not-started',
+    priority: 'high',
+    projectId: 'project-1',
+    missionId: null,
+    assignee: null,
+    tags: null,
+    dueDate: null,
+    metadata: null,
+    createdAt: '2026-08-01T00:00:00.000Z',
+    updatedAt: '2026-08-01T00:00:00.000Z',
+  },
+];
+
+function acceptedTaskSnapshot(
+  overrides: Partial<FreshSnapshot<typeof taskPayload>> = {},
+): FreshSnapshot<typeof taskPayload> {
+  const result = acceptSnapshot({
+    value: taskPayload,
+    validate: validateTaskCollection,
+    previous: null,
+    policy,
+    source: 'gateway:/api/tasks',
+    now: NOW,
+  });
+  if (result.outcome !== 'accepted') {
+    throw new Error(`fixture setup failed: ${result.reason}`);
+  }
+  return { ...result.snapshot, ...overrides };
+}
+
+describe('computeFreshness', () => {
+  it('treats a missing snapshot as unavailable, never as an empty healthy collection', () => {
+    expect(computeFreshness({ snapshot: null, policy, now: NOW })).toBe('unavailable');
+  });
+
+  it('returns current for a fresh verified snapshot regardless of data emptiness', () => {
+    const empty = acceptSnapshot({
+      value: [],
+      validate: validateTaskCollection,
+      previous: null,
+      policy,
+      source: 'gateway:/api/tasks',
+      now: NOW,
+    });
+    if (empty.outcome !== 'accepted') throw new Error('expected acceptance');
+    expect(computeFreshness({ snapshot: empty.snapshot, policy, now: NOW })).toBe('current');
+  });
+
+  it('degrades to stale once the snapshot ages past staleAfterMs', () => {
+    const snapshot = acceptedTaskSnapshot();
+    expect(computeFreshness({ snapshot, policy, now: NOW + 60_001 })).toBe('stale');
+    expect(computeFreshness({ snapshot, policy, now: NOW + 59_999 })).toBe('current');
+  });
+
+  it('degrades to stale when the latest revalidation failed', () => {
+    const snapshot = acceptedTaskSnapshot();
+    expect(computeFreshness({ snapshot, policy, now: NOW, degraded: true })).toBe('stale');
+  });
+});
+
+describe('mutation guard', () => {
+  it('permits mutations only on current data', () => {
+    expect(canMutate('current')).toBe(true);
+    for (const state of ['stale', 'partial', 'unknown', 'unavailable'] as const) {
+      expect(canMutate(state)).toBe(false);
+    }
+  });
+
+  it('refuses mutations on non-current data via assertMutable', () => {
+    expect(() => assertMutable('current')).not.toThrow();
+    for (const state of ['stale', 'partial', 'unknown', 'unavailable'] as const) {
+      let thrown: unknown;
+      try {
+        assertMutable(state);
+      } catch (caught) {
+        thrown = caught;
+      }
+      expect(thrown).toBeInstanceOf(StaleMutationError);
+      expect(thrown).toBeInstanceOf(Error);
+      if (thrown instanceof StaleMutationError) {
+        expect(thrown.name).toBe('StaleMutationError');
+        expect(thrown.freshness).toBe(state);
+        expect(thrown.message).toContain(state);
+        expect(thrown.message).toContain('revalidat');
+      }
+    }
+  });
+});
+
+describe('acceptSnapshot', () => {
+  it('accepts a valid payload with provenance', () => {
+    const result = acceptSnapshot({
+      value: taskPayload,
+      validate: validateTaskCollection,
+      previous: null,
+      policy,
+      source: 'gateway:/api/tasks',
+      now: NOW,
+    });
+    expect(result.outcome).toBe('accepted');
+    if (result.outcome !== 'accepted') return;
+    expect(result.snapshot.source).toBe('gateway:/api/tasks');
+    expect(result.snapshot.version).toBe(1);
+    expect(result.snapshot.fetchedAt).toBe(NOW);
+    expect(result.snapshot.data).toEqual(taskPayload);
+  });
+
+  it('invalidates a schema-mismatched payload instead of rendering it', () => {
+    const result = acceptSnapshot({
+      value: { not: 'an array' },
+      validate: validateTaskCollection,
+      previous: acceptedTaskSnapshot(),
+      policy,
+      source: 'gateway:/api/tasks',
+      now: NOW,
+    });
+    expect(result).toEqual({ outcome: 'invalidated', reason: 'schema-mismatch' });
+    expect(invalidationReasonLabels['schema-mismatch']).toContain('schema');
+  });
+
+  it('invalidates cross-workspace payloads', () => {
+    const userOne = acceptSnapshot({
+      value: [
+        {
+          id: 'p1',
+          name: 'P1',
+          description: null,
+          status: 'active',
+          userId: 'user-1',
+          metadata: null,
+          createdAt: '2026-08-01T00:00:00.000Z',
+          updatedAt: '2026-08-01T00:00:00.000Z',
+        },
+      ],
+      validate: validateProjectCollection,
+      previous: null,
+      policy,
+      source: 'gateway:/api/projects',
+      now: NOW,
+    });
+    if (userOne.outcome !== 'accepted') throw new Error('expected acceptance');
+
+    const switched = acceptSnapshot({
+      value: [
+        {
+          id: 'p9',
+          name: 'P9',
+          description: null,
+          status: 'active',
+          userId: 'user-2',
+          metadata: null,
+          createdAt: '2026-08-01T00:00:00.000Z',
+          updatedAt: '2026-08-01T00:00:00.000Z',
+        },
+      ],
+      validate: validateProjectCollection,
+      previous: userOne.snapshot,
+      policy,
+      source: 'gateway:/api/projects',
+      now: NOW,
+    });
+    expect(switched).toEqual({ outcome: 'invalidated', reason: 'cross-workspace' });
+  });
+
+  it('keeps the previous workspace for collections with no intrinsic identity', () => {
+    const userOne = acceptSnapshot({
+      value: [
+        {
+          id: 'p1',
+          name: 'P1',
+          description: null,
+          status: 'active',
+          userId: 'user-1',
+          metadata: null,
+          createdAt: '2026-08-01T00:00:00.000Z',
+          updatedAt: '2026-08-01T00:00:00.000Z',
+        },
+      ],
+      validate: validateProjectCollection,
+      previous: null,
+      policy,
+      source: 'gateway:/api/projects',
+      now: NOW,
+    });
+    if (userOne.outcome !== 'accepted') throw new Error('expected acceptance');
+
+    // Empty list after the user deleted every project: no identity to check,
+    // so the verified scope is retained and the empty state stays healthy.
+    const emptied = acceptSnapshot({
+      value: [],
+      validate: validateProjectCollection,
+      previous: userOne.snapshot,
+      policy,
+      source: 'gateway:/api/projects',
+      now: NOW,
+    });
+    expect(emptied.outcome).toBe('accepted');
+    if (emptied.outcome === 'accepted') {
+      expect(emptied.snapshot.data).toEqual([]);
+      expect(emptied.snapshot.workspace).toBe('user-1');
+    }
+  });
+
+  it('invalidates version regressions', () => {
+    const previous = acceptedTaskSnapshot({ version: 7 });
+    const regressed = acceptSnapshot({
+      value: taskPayload,
+      validate: validateTaskCollection,
+      previous,
+      policy,
+      source: 'gateway:/api/tasks',
+      now: NOW,
+      incomingVersion: 3,
+    });
+    expect(regressed).toEqual({ outcome: 'invalidated', reason: 'version-regression' });
+
+    const newerSchema = acceptedTaskSnapshot({ schemaVersion: 4 });
+    const downgradedClient = acceptSnapshot({
+      value: taskPayload,
+      validate: validateTaskCollection,
+      previous: newerSchema,
+      policy: { ...policy, schemaVersion: 2 },
+      source: 'gateway:/api/tasks',
+      now: NOW,
+    });
+    expect(downgradedClient).toEqual({ outcome: 'invalidated', reason: 'version-regression' });
+  });
+
+  it('increments the version monotonically across accepted snapshots', () => {
+    const first = acceptedTaskSnapshot();
+    const second = acceptSnapshot({
+      value: taskPayload,
+      validate: validateTaskCollection,
+      previous: first,
+      policy,
+      source: 'gateway:/api/tasks',
+      now: NOW,
+    });
+    expect(second.outcome).toBe('accepted');
+    if (second.outcome === 'accepted') {
+      expect(second.snapshot.version).toBe(first.version + 1);
+    }
+  });
+});
+
+describe('combineFreshness', () => {
+  it('gates the surface on the primary collection', () => {
+    expect(combineFreshness('unavailable', ['current'])).toBe('unavailable');
+    expect(combineFreshness('unknown', ['current'])).toBe('unknown');
+    expect(combineFreshness('current', [])).toBe('current');
+  });
+
+  it('degrades to partial when a secondary is unavailable', () => {
+    expect(combineFreshness('current', ['current', 'unavailable'])).toBe('partial');
+  });
+
+  it('degrades to unknown while a secondary is still loading', () => {
+    expect(combineFreshness('current', ['unknown'])).toBe('unknown');
+  });
+
+  it('degrades to stale when any collection is stale', () => {
+    expect(combineFreshness('current', ['stale'])).toBe('stale');
+    expect(combineFreshness('stale', ['current'])).toBe('stale');
+  });
+
+  it('propagates partial secondaries', () => {
+    expect(combineFreshness('current', ['partial'])).toBe('partial');
+  });
+});
+
+describe('computeDigest', () => {
+  it('is stable across key order and changes with data', () => {
+    const a = computeDigest({ x: 1, y: [1, 2] });
+    const b = computeDigest({ y: [1, 2], x: 1 });
+    expect(a).toBe(b);
+    expect(computeDigest({ x: 1, y: [1, 3] })).not.toBe(a);
+  });
+});
+
+describe('verdictValue', () => {
+  it('returns the value only for verified inputs', () => {
+    expect(verdictValue(true, '5')).toBe('5');
+    expect(verdictValue(false, '5')).toBe(UNKNOWN_VERDICT);
+    expect(verdictValue(false, '5')).not.toBe('5');
+  });
+});
+
+describe('formatAge', () => {
+  it('labels age in human terms', () => {
+    expect(formatAge(NOW, NOW)).toBe('just now');
+    expect(formatAge(NOW, NOW + 15_000)).toBe('under a minute ago');
+    expect(formatAge(NOW, NOW + 120_000)).toBe('2m ago');
+    expect(formatAge(NOW, NOW + 3 * 3_600_000)).toBe('3h ago');
+    expect(formatAge(NOW, NOW + 2 * 86_400_000)).toBe('2d ago');
+  });
+});

apps/web/src/lib/freshness/model.ts

@@ -0,0 +1,261 @@
+/**
+ * Typed freshness model for gateway-fetched collections (RI-5-001).
+ *
+ * A failed or stale fetch must never be indistinguishable from an empty
+ * healthy collection. Every fetched surface carries an explicit freshness
+ * state, a verified snapshot identity (source, workspace, version, age), and
+ * a mutation guard that refuses state-changing operations unless the data is
+ * verified current.
+ */
+
+/** Freshness states for fetched data. Never inferred from emptiness. */
+export type FreshnessState = 'current' | 'stale' | 'partial' | 'unknown' | 'unavailable';
+
+/**
+ * Reasons a snapshot is invalidated. An invalidated snapshot is treated as
+ * unavailable and is never rendered as current.
+ */
+export type InvalidationReason =
+  | 'cache-corruption'
+  | 'cross-workspace'
+  | 'schema-mismatch'
+  | 'version-regression';
+
+/** Human-readable labels for invalidation reasons (UI + error messages). */
+export const invalidationReasonLabels: Record<InvalidationReason, string> = {
+  'cache-corruption': 'cached snapshot failed integrity checks',
+  'cross-workspace': 'data belongs to a different workspace',
+  'schema-mismatch': 'response did not match the expected schema',
+  'version-regression': 'snapshot version regressed below the accepted version',
+};
+
+/** A verified snapshot of fetched data with full provenance. */
+export interface FreshSnapshot<T> {
+  readonly data: T;
+  /** Source identity of the fetch, e.g. `gateway:/api/tasks`. */
+  readonly source: string;
+  /** Workspace scope the data belongs to. */
+  readonly workspace: string;
+  /** Monotonic snapshot sequence number for this surface. */
+  readonly version: number;
+  /** Schema version of the validator that accepted this snapshot. */
+  readonly schemaVersion: number;
+  /** Epoch ms at which the data was verified. */
+  readonly fetchedAt: number;
+  /** Integrity digest of `data`, used to detect cache corruption. */
+  readonly digest: string;
+}
+
+/** Provenance label rendered next to last-known data. */
+export interface FreshnessLabel {
+  readonly source: string;
+  readonly version: number;
+  readonly fetchedAt: number;
+}
+
+/** Policy governing freshness for a surface. */
+export interface FreshnessPolicy {
+  /** Active workspace scope. Snapshots from other scopes are invalidated. */
+  readonly workspace: string;
+  /** Schema version of the current validator. */
+  readonly schemaVersion: number;
+  /** Age after which a verified snapshot degrades from current to stale. */
+  readonly staleAfterMs: number;
+}
+
+export const DEFAULT_FRESHNESS_POLICY: FreshnessPolicy = {
+  workspace: 'default',
+  schemaVersion: 1,
+  staleAfterMs: 60_000,
+};
+
+/** Payload returned by a successful schema validation. */
+export interface FreshPayload<T> {
+  readonly data: T;
+  /**
+   * Workspace identity extracted from the payload itself when the collection
+   * carries one (e.g. a uniform `userId` on projects). `null` when the
+   * collection has no intrinsic workspace identity.
+   */
+  readonly workspace: string | null;
+}
+
+/** Error thrown when a mutation is attempted on non-current data. */
+export class StaleMutationError extends Error {
+  readonly freshness: FreshnessState;
+
+  constructor(freshness: FreshnessState) {
+    super(`Refused mutation on ${freshness} data: revalidation is required before mutating.`);
+    this.name = 'StaleMutationError';
+    this.freshness = freshness;
+  }
+}
+
+/** Stable JSON digest used for snapshot integrity checks. */
+export function computeDigest(value: unknown): string {
+  // FNV-1a 32-bit over the stable JSON serialization. This is an integrity
+  // check against corruption, not a cryptographic guarantee.
+  let hash = 0x811c9dc5;
+  for (const byte of stableStringify(value)) {
+    hash ^= byte.charCodeAt(0);
+    hash = Math.imul(hash, 0x01000193) >>> 0;
+  }
+  return hash.toString(16).padStart(8, '0');
+}
+
+function stableStringify(value: unknown): string {
+  return serialize(value);
+}
+
+function serialize(value: unknown): string {
+  if (value === null || typeof value !== 'object') return JSON.stringify(value) ?? 'null';
+  if (Array.isArray(value)) return `[${value.map(serialize).join(',')}]`;
+  const entries = Object.entries(value as Record<string, unknown>)
+    .filter(([, item]) => item !== undefined)
+    .sort(([left], [right]) => (left < right ? -1 : left > right ? 1 : 0))
+    .map(([key, item]) => `${JSON.stringify(key)}:${serialize(item)}`);
+  return `{${entries.join(',')}}`;
+}
+
+export type AcceptSnapshotResult<T> =
+  | { readonly outcome: 'accepted'; readonly snapshot: FreshSnapshot<T> }
+  | { readonly outcome: 'invalidated'; readonly reason: InvalidationReason };
+
+export interface AcceptSnapshotOptions<T> {
+  /** Raw fetched value (untrusted JSON). */
+  readonly value: unknown;
+  /** Schema validator; returns `null` when the value does not match. */
+  readonly validate: (value: unknown) => FreshPayload<T> | null;
+  /** Previously accepted snapshot for this surface, if any. */
+  readonly previous: FreshSnapshot<T> | null;
+  readonly policy: FreshnessPolicy;
+  readonly source: string;
+  /**
+   * Version carried by the incoming payload when the transport exposes one.
+   * Must not regress below the accepted snapshot's version.
+   */
+  readonly incomingVersion?: number;
+  readonly now: number;
+}
+
+/**
+ * Validate and accept a fetched value as a snapshot, or invalidate it.
+ *
+ * Invalidation rules (each treated as unavailable, never rendered current):
+ * - schema mismatch: the payload fails validation
+ * - cross-workspace: the payload's workspace differs from the verified one
+ * - version regression: payload/schema version is below the accepted one
+ */
+export function acceptSnapshot<T>(options: AcceptSnapshotOptions<T>): AcceptSnapshotResult<T> {
+  const payload = options.validate(options.value);
+  if (payload === null) {
+    return { outcome: 'invalidated', reason: 'schema-mismatch' };
+  }
+
+  // Workspace identity: the payload's own scope wins; a collection with no
+  // intrinsic identity (e.g. an empty list after every project was deleted)
+  // keeps the previously verified scope rather than resetting to the policy
+  // default, so a legitimately empty response is not mistaken for a scope
+  // change.
+  const workspace = payload.workspace ?? options.previous?.workspace ?? options.policy.workspace;
+  if (options.previous !== null && options.previous.workspace !== workspace) {
+    return { outcome: 'invalidated', reason: 'cross-workspace' };
+  }
+  if (options.previous !== null && options.policy.schemaVersion < options.previous.schemaVersion) {
+    return { outcome: 'invalidated', reason: 'version-regression' };
+  }
+  if (
+    options.incomingVersion !== undefined &&
+    options.previous !== null &&
+    options.incomingVersion < options.previous.version
+  ) {
+    return { outcome: 'invalidated', reason: 'version-regression' };
+  }
+
+  const snapshot: FreshSnapshot<T> = {
+    data: payload.data,
+    source: options.source,
+    workspace,
+    version: options.incomingVersion ?? (options.previous?.version ?? 0) + 1,
+    schemaVersion: options.policy.schemaVersion,
+    fetchedAt: options.now,
+    digest: computeDigest(payload.data),
+  };
+  return { outcome: 'accepted', snapshot };
+}
+
+export interface ComputeFreshnessOptions {
+  readonly snapshot: FreshSnapshot<unknown> | null;
+  readonly policy: FreshnessPolicy;
+  readonly now: number;
+  /**
+   * True when the snapshot cannot be trusted as current regardless of age:
+   * the latest revalidation failed, or the snapshot was restored from cache
+   * and has not been verified by a fetch in this session.
+   */
+  readonly degraded?: boolean;
+}
+
+/**
+ * Compute the freshness state of a snapshot. A missing snapshot is
+ * `unavailable` (never "empty and healthy"); a degraded or aged snapshot is
+ * `stale` (situational awareness only).
+ */
+export function computeFreshness(options: ComputeFreshnessOptions): FreshnessState {
+  const { snapshot, policy, now, degraded = false } = options;
+  if (snapshot === null) return 'unavailable';
+  if (degraded) return 'stale';
+  if (now - snapshot.fetchedAt > policy.staleAfterMs) return 'stale';
+  return 'current';
+}
+
+/** Only verified-current data may back a state-changing action. */
+export function canMutate(state: FreshnessState): boolean {
+  return state === 'current';
+}
+
+/** Defense in depth: reject the mutation call itself on non-current data. */
+export function assertMutable(state: FreshnessState): void {
+  if (!canMutate(state)) {
+    throw new StaleMutationError(state);
+  }
+}
+
+/**
+ * Combine freshness across a multi-collection surface (primary + secondaries).
+ * The primary collection gates the surface: unknown while it loads,
+ * unavailable when it fails. Missing secondaries degrade the surface to
+ * `partial`; aged collections degrade it to `stale`.
+ */
+export function combineFreshness(
+  primary: FreshnessState,
+  secondaries: readonly FreshnessState[],
+): FreshnessState {
+  if (primary === 'unavailable') return 'unavailable';
+  if (primary === 'unknown') return 'unknown';
+  if (secondaries.includes('unavailable')) return 'partial';
+  if (secondaries.includes('unknown')) return 'unknown';
+  if (secondaries.includes('stale') || primary === 'stale') return 'stale';
+  if (secondaries.includes('partial')) return 'partial';
+  return 'current';
+}
+
+/** Render-safe age label for snapshot provenance. */
+export function formatAge(fetchedAt: number, now: number): string {
+  const ageMs = Math.max(0, now - fetchedAt);
+  if (ageMs < 10_000) return 'just now';
+  const minutes = Math.floor(ageMs / 60_000);
+  if (minutes < 1) return 'under a minute ago';
+  if (minutes < 60) return `${minutes}m ago`;
+  const hours = Math.floor(minutes / 60);
+  if (hours < 24) return `${hours}h ago`;
+  const days = Math.floor(hours / 24);
+  return `${days}d ago`;
+}
+
+/** Derived verdict placeholder for non-current inputs — never a green value. */
+export const UNKNOWN_VERDICT = '?';
+
+export function verdictValue(verified: boolean, value: string): string {
+  return verified ? value : UNKNOWN_VERDICT;
+}

apps/web/src/lib/freshness/snapshot-cache.spec.ts

@@ -0,0 +1,197 @@
+import { afterEach, beforeEach, describe, expect, it } from 'vitest';
+import { acceptSnapshot, DEFAULT_FRESHNESS_POLICY } from './model';
+import { clearSnapshotCache, readSnapshotCache, writeSnapshotCache } from './snapshot-cache';
+import { validateProjectCollection, validateTaskCollection } from './validators';
+import { projectFixtures, taskFixtures } from '@/spa/pages/page-fixtures';
+import type { Project, Task } from '@/lib/types';
+
+const KEY = 'test:tasks';
+const NOW = 1_800_000_000_000;
+const policy = { ...DEFAULT_FRESHNESS_POLICY, staleAfterMs: 60_000 };
+
+function storedTaskSnapshot() {
+  const result = acceptSnapshot({
+    value: taskFixtures,
+    validate: validateTaskCollection,
+    previous: null,
+    policy,
+    source: 'gateway:/api/tasks',
+    now: NOW,
+  });
+  if (result.outcome !== 'accepted') throw new Error('fixture setup failed');
+  return result.snapshot;
+}
+
+function storedProjectSnapshot() {
+  const result = acceptSnapshot({
+    value: projectFixtures,
+    validate: validateProjectCollection,
+    previous: null,
+    policy,
+    source: 'gateway:/api/projects',
+    now: NOW,
+  });
+  if (result.outcome !== 'accepted') throw new Error('fixture setup failed');
+  return result.snapshot;
+}
+
+function readTasks() {
+  return readSnapshotCache({
+    key: KEY,
+    workspace: policy.workspace,
+    policy,
+    validate: validateTaskCollection,
+  });
+}
+
+/** Write an arbitrary value directly at the raw cache slot. */
+function writeRaw(key: string, value: unknown): void {
+  sessionStorage.setItem(`mosaic:freshness:v1:${key}`, JSON.stringify(value));
+}
+
+/** Parse and re-write the stored entry (for tampering with internals). */
+function tamperStored<T>(key: string, mutate: (stored: T) => void): void {
+  const parsed = JSON.parse(sessionStorage.getItem(`mosaic:freshness:v1:${key}`) ?? '{}') as T;
+  mutate(parsed);
+  writeRaw(key, parsed);
+}
+
+beforeEach(() => {
+  sessionStorage.clear();
+});
+
+afterEach(() => {
+  sessionStorage.clear();
+});
+
+describe('readSnapshotCache', () => {
+  it('misses when nothing is stored', () => {
+    expect(readTasks()).toEqual({ outcome: 'miss' });
+  });
+
+  it('hits for a well-formed entry and preserves provenance', () => {
+    const snapshot = storedTaskSnapshot();
+    writeSnapshotCache(KEY, snapshot);
+
+    const result = readTasks();
+    expect(result.outcome).toBe('hit');
+    if (result.outcome === 'hit') {
+      expect(result.snapshot.data).toEqual(taskFixtures);
+      expect(result.snapshot.source).toBe('gateway:/api/tasks');
+      expect(result.snapshot.version).toBe(snapshot.version);
+      expect(result.snapshot.fetchedAt).toBe(snapshot.fetchedAt);
+      expect(result.snapshot.workspace).toBe(snapshot.workspace);
+    }
+  });
+
+  it('invalidates unparsable entries as cache corruption', () => {
+    sessionStorage.setItem(`mosaic:freshness:v1:${KEY}`, '{not json');
+    expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'cache-corruption' });
+  });
+
+  it('invalidates structurally wrong entries as cache corruption', () => {
+    const malformed: unknown[] = [
+      'nested but not a snapshot',
+      { data: taskFixtures }, // missing provenance fields
+      {
+        data: taskFixtures,
+        source: 1,
+        workspace: 'w',
+        version: 1,
+        schemaVersion: 1,
+        fetchedAt: 1,
+        digest: 'x',
+      },
+      null,
+      17,
+    ];
+    for (const entry of malformed) {
+      writeRaw(KEY, entry);
+      expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'cache-corruption' });
+    }
+  });
+
+  it('invalidates digest mismatches as cache corruption (tampered data)', () => {
+    writeSnapshotCache(KEY, storedTaskSnapshot());
+    tamperStored<{ data: Task[] }>(KEY, (stored) => {
+      stored.data = [...stored.data, { ...stored.data[0]!, id: 'injected-task' }];
+    });
+    expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'cache-corruption' });
+  });
+
+  it('invalidates entries scoped to another workspace', () => {
+    const snapshot = storedTaskSnapshot();
+    writeSnapshotCache(KEY, { ...snapshot, workspace: 'someone-else' });
+    expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'cross-workspace' });
+  });
+
+  it('invalidates entries written by a newer schema as a version regression', () => {
+    const snapshot = storedTaskSnapshot();
+    writeSnapshotCache(KEY, { ...snapshot, schemaVersion: policy.schemaVersion + 1 });
+    expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'version-regression' });
+  });
+
+  it('invalidates entries whose data no longer validates (schema mismatch)', () => {
+    writeSnapshotCache(KEY, storedTaskSnapshot());
+    tamperStored<{ data: unknown }>(KEY, (stored) => {
+      stored.data = { malformed: true };
+    });
+    expect(readTasks()).toEqual({ outcome: 'invalidated', reason: 'schema-mismatch' });
+  });
+
+  it('never reports a corrupted raw entry as a hit (negative control)', () => {
+    for (const raw of ['{oops', 'null', '"string"', '[]', '12']) {
+      sessionStorage.setItem(`mosaic:freshness:v1:${KEY}`, raw);
+      const result = readTasks();
+      expect(result.outcome).not.toBe('hit');
+      expect(result.outcome).toBe('invalidated');
+    }
+  });
+
+  it('scopes project collections by their workspace identity', () => {
+    const snapshot = storedProjectSnapshot();
+    writeSnapshotCache('test:projects', snapshot);
+
+    const sameScope = readSnapshotCache({
+      key: 'test:projects',
+      workspace: 'user-1',
+      policy,
+      validate: validateProjectCollection,
+    });
+    expect(sameScope.outcome).toBe('hit');
+
+    const foreignScope = readSnapshotCache({
+      key: 'test:projects',
+      workspace: 'user-2',
+      policy,
+      validate: validateProjectCollection,
+    });
+    expect(foreignScope).toEqual({ outcome: 'invalidated', reason: 'cross-workspace' });
+  });
+});
+
+describe('writeSnapshotCache round-trip', () => {
+  it('round-trips an accepted project snapshot', () => {
+    const snapshot = storedProjectSnapshot();
+    writeSnapshotCache('test:projects', snapshot);
+    const result = readSnapshotCache({
+      key: 'test:projects',
+      workspace: snapshot.workspace,
+      policy,
+      validate: validateProjectCollection,
+    });
+    expect(result.outcome).toBe('hit');
+    if (result.outcome === 'hit') {
+      expect(result.snapshot.data).toEqual(projectFixtures as Project[]);
+    }
+  });
+});
+
+describe('clearSnapshotCache', () => {
+  it('drops the entry so the next read misses', () => {
+    writeSnapshotCache(KEY, storedTaskSnapshot());
+    expect(readTasks().outcome).toBe('hit');
+    clearSnapshotCache(KEY);
+    expect(readTasks()).toEqual({ outcome: 'miss' });
+  });
+});

apps/web/src/lib/freshness/snapshot-cache.ts

@@ -0,0 +1,154 @@
+import {
+  computeDigest,
+  type FreshPayload,
+  type FreshSnapshot,
+  type FreshnessPolicy,
+  type InvalidationReason,
+} from './model';
+
+/**
+ * Session-scoped last-known snapshot cache (RI-5-001).
+ *
+ * Restored snapshots are situational awareness only: they surface as `stale`
+ * until a fetch re-verifies them. A cache entry that is corrupted, belongs to
+ * another workspace, was written by a newer schema, or no longer validates is
+ * invalidated (treated as unavailable, never rendered as current).
+ */
+
+const CACHE_PREFIX = 'mosaic:freshness:v1';
+
+interface StoredSnapshot {
+  data: unknown;
+  source: string;
+  workspace: string;
+  version: number;
+  schemaVersion: number;
+  fetchedAt: number;
+  digest: string;
+}
+
+export type SnapshotCacheRead<T> =
+  | { readonly outcome: 'hit'; readonly snapshot: FreshSnapshot<T> }
+  | { readonly outcome: 'miss' }
+  | { readonly outcome: 'invalidated'; readonly reason: InvalidationReason };
+
+export interface ReadSnapshotCacheOptions<T> {
+  readonly key: string;
+  readonly workspace: string;
+  readonly policy: FreshnessPolicy;
+  readonly validate: (value: unknown) => FreshPayload<T> | null;
+}
+
+function cacheKey(key: string): string {
+  return `${CACHE_PREFIX}:${key}`;
+}
+
+function isStoredSnapshot(value: unknown): value is StoredSnapshot {
+  if (typeof value !== 'object' || value === null) return false;
+  const candidate = value as Record<string, unknown>;
+  return (
+    typeof candidate['data'] === 'object' &&
+    candidate['data'] !== null &&
+    typeof candidate['source'] === 'string' &&
+    typeof candidate['workspace'] === 'string' &&
+    typeof candidate['version'] === 'number' &&
+    typeof candidate['schemaVersion'] === 'number' &&
+    typeof candidate['fetchedAt'] === 'number' &&
+    typeof candidate['digest'] === 'string'
+  );
+}
+
+function getStorage(): Storage | null {
+  try {
+    return globalThis.sessionStorage ?? null;
+  } catch {
+    return null;
+  }
+}
+
+/**
+ * Restore a cached snapshot under the active workspace scope. Every failure
+ * mode maps to an explicit invalidation reason or a miss — never to data
+ * that renders as current.
+ */
+export function readSnapshotCache<T>(options: ReadSnapshotCacheOptions<T>): SnapshotCacheRead<T> {
+  const storage = getStorage();
+  if (storage === null) return { outcome: 'miss' };
+
+  let raw: string | null;
+  try {
+    raw = storage.getItem(cacheKey(options.key));
+  } catch {
+    return { outcome: 'miss' };
+  }
+  if (raw === null) return { outcome: 'miss' };
+
+  let parsed: unknown;
+  try {
+    parsed = JSON.parse(raw);
+  } catch {
+    return { outcome: 'invalidated', reason: 'cache-corruption' };
+  }
+  if (!isStoredSnapshot(parsed)) {
+    return { outcome: 'invalidated', reason: 'cache-corruption' };
+  }
+  if (parsed.workspace !== options.workspace) {
+    return { outcome: 'invalidated', reason: 'cross-workspace' };
+  }
+  if (parsed.schemaVersion > options.policy.schemaVersion) {
+    // Written by a newer build than the running client: version regression.
+    return { outcome: 'invalidated', reason: 'version-regression' };
+  }
+
+  const payload = options.validate(parsed.data);
+  if (payload === null) {
+    return { outcome: 'invalidated', reason: 'schema-mismatch' };
+  }
+  if (computeDigest(payload.data) !== parsed.digest) {
+    return { outcome: 'invalidated', reason: 'cache-corruption' };
+  }
+
+  return {
+    outcome: 'hit',
+    snapshot: {
+      data: payload.data,
+      source: parsed.source,
+      workspace: parsed.workspace,
+      version: parsed.version,
+      schemaVersion: parsed.schemaVersion,
+      fetchedAt: parsed.fetchedAt,
+      digest: parsed.digest,
+    },
+  };
+}
+
+/** Persist a verified snapshot. Failures are non-fatal (cache is best-effort). */
+export function writeSnapshotCache<T>(key: string, snapshot: FreshSnapshot<T>): void {
+  const storage = getStorage();
+  if (storage === null) return;
+  const stored: StoredSnapshot = {
+    data: snapshot.data,
+    source: snapshot.source,
+    workspace: snapshot.workspace,
+    version: snapshot.version,
+    schemaVersion: snapshot.schemaVersion,
+    fetchedAt: snapshot.fetchedAt,
+    digest: snapshot.digest,
+  };
+  try {
+    storage.setItem(cacheKey(key), JSON.stringify(stored));
+  } catch {
+    // Quota or serialization failures simply skip caching.
+  }
+}
+
+/** Drop a cached snapshot (used when a surface invalidates its cache entry). */
+export function clearSnapshotCache(key: string): void {
+  const storage = getStorage();
+  if (storage === null) return;
+  try {
+    storage.removeItem(cacheKey(key));
+  } catch {
+    // Ignorable: a wedged storage entry is detected as corruption on read.
+  }
+}

apps/web/src/lib/freshness/use-fresh-collection.spec.tsx

@@ -0,0 +1,372 @@
+import { act } from 'react';
+import { createRoot, type Root } from 'react-dom/client';
+import { afterEach, beforeAll, beforeEach, describe, expect, it, vi } from 'vitest';
+import type { Task } from '@/lib/types';
+import { acceptSnapshot, StaleMutationError, DEFAULT_FRESHNESS_POLICY } from './model';
+import type { FreshnessFailure } from './use-fresh-collection';
+import {
+  describeFailure,
+  useFreshCollection,
+  type FreshCollection,
+  type UseFreshCollectionOptions,
+} from './use-fresh-collection';
+import { validateProjectCollection, validateTaskCollection } from './validators';
+import { projectFixtures, taskFixtures } from '@/spa/pages/page-fixtures';
+
+/**
+ * Failure-matrix coverage for the freshness seam (RI-5-001): network failure,
+ * auth failure, malformed response, cache corruption, stale age, schema
+ * mismatch, cross-workspace, recovery, and stale-action rejection — with
+ * negative controls proving no case yields current data or an enabled
+ * mutation.
+ */
+
+const NOW = 1_800_000_000_000;
+
+interface Deferred<T> {
+  promise: Promise<T>;
+  resolve: (value: T) => void;
+  reject: (reason?: unknown) => void;
+}
+
+function createDeferred<T>(): Deferred<T> {
+  let resolve!: (value: T) => void;
+  let reject!: (reason?: unknown) => void;
+  const promise = new Promise<T>((res, rej) => {
+    resolve = res;
+    reject = rej;
+  });
+  return { promise, resolve, reject };
+}
+
+let root: Root | null = null;
+let container: HTMLDivElement;
+let latest: FreshCollection<Task[]> | null = null;
+
+function Probe({
+  options,
+}: {
+  options: UseFreshCollectionOptions<Task[]>;
+}): React.ReactElement | null {
+  latest = useFreshCollection<Task[]>(options);
+  return null;
+}
+
+beforeAll(() => {
+  Object.defineProperty(globalThis, 'IS_REACT_ACT_ENVIRONMENT', {
+    configurable: true,
+    value: true,
+  });
+});
+
+beforeEach(() => {
+  sessionStorage.clear();
+});
+
+afterEach(async () => {
+  await act(async () => {
+    root?.unmount();
+  });
+  document.body.replaceChildren();
+  root = null;
+  latest = null;
+  sessionStorage.clear();
+  vi.restoreAllMocks();
+});
+
+async function renderCollection(
+  options: UseFreshCollectionOptions<Task[]>,
+): Promise<FreshCollection<Task[]>> {
+  container = document.createElement('div');
+  document.body.append(container);
+  root = createRoot(container);
+  await act(async () => {
+    root?.render(<Probe options={options} />);
+  });
+  if (latest === null) throw new Error('hook did not run');
+  return latest;
+}
+
+function taskOptions(
+  overrides: Partial<UseFreshCollectionOptions<Task[]>> = {},
+): UseFreshCollectionOptions<Task[]> {
+  return {
+    source: 'gateway:/api/tasks',
+    fetcher: () => Promise.resolve(taskFixtures),
+    validate: validateTaskCollection,
+    cacheKey: 'tasks',
+    clock: () => NOW,
+    ...overrides,
+  };
+}
+
+function authError(statusCode: number): Error & { statusCode: number } {
+  return Object.assign(new Error(`Request failed with ${statusCode}`), { statusCode });
+}
+
+function seedCache(key: string): number {
+  const result = acceptSnapshot({
+    value: taskFixtures,
+    validate: validateTaskCollection,
+    previous: null,
+    policy: DEFAULT_FRESHNESS_POLICY,
+    source: 'gateway:/api/tasks',
+    now: NOW,
+  });
+  if (result.outcome !== 'accepted') throw new Error('fixture setup failed');
+  sessionStorage.setItem(`mosaic:freshness:v1:${key}`, JSON.stringify({ ...result.snapshot }));
+  return result.snapshot.version;
+}
+
+describe('useFreshCollection failure matrix', () => {
+  it('is unknown (not empty) while the first validation is in flight', async () => {
+    const deferred = createDeferred<Task[]>();
+    const collection = await renderCollection(taskOptions({ fetcher: () => deferred.promise }));
+
+    expect(collection.freshness).toBe('unknown');
+    expect(collection.validating).toBe(true);
+    expect(collection.data).toBeNull();
+    expect(collection.canMutate).toBe(false);
+
+    await act(async () => {
+      deferred.resolve(taskFixtures);
+      await deferred.promise;
+    });
+  });
+
+  it('becomes current with provenance after a verified fetch', async () => {
+    const collection = await renderCollection(taskOptions());
+
+    expect(collection.freshness).toBe('current');
+    expect(collection.data).toEqual(taskFixtures);
+    expect(collection.snapshot?.source).toBe('gateway:/api/tasks');
+    expect(collection.snapshot?.version).toBe(1);
+    expect(collection.failure).toBeNull();
+    expect(collection.canMutate).toBe(true);
+    // Verified snapshot is persisted for last-known restore.
+    expect(sessionStorage.getItem('mosaic:freshness:v1:tasks')).toBeTruthy();
+  });
+
+  it('treats a network failure as unavailable — never an empty healthy collection', async () => {
+    const collection = await renderCollection(
+      taskOptions({ fetcher: () => Promise.reject(new Error('network down')) }),
+    );
+
+    expect(collection.freshness).toBe('unavailable');
+    expect(collection.data).toBeNull();
+    expect(collection.failure).toEqual({ kind: 'fetch', message: 'network down' });
+    expect(collection.canMutate).toBe(false);
+    expect(describeFailure(collection.failure)).toBe('network down');
+  });
+
+  it('treats an auth failure as unavailable and drops the last-known snapshot', async () => {
+    let call = 0;
+    const collection = await renderCollection(
+      taskOptions({
+        fetcher: () => {
+          call += 1;
+          return call === 1 ? Promise.resolve(taskFixtures) : Promise.reject(authError(401));
+        },
+      }),
+    );
+    expect(collection.freshness).toBe('current');
+
+    await act(async () => {
+      await collection.revalidate();
+    });
+
+    expect(latest?.freshness).toBe('unavailable');
+    expect(latest?.data).toBeNull();
+    expect(latest?.failure?.kind).toBe('fetch');
+    // The previous user's data must not linger in the session cache.
+    expect(sessionStorage.getItem('mosaic:freshness:v1:tasks')).toBeNull();
+  });
+
+  it('invalidates a malformed response as a schema mismatch', async () => {
+    const collection = await renderCollection(
+      taskOptions({ fetcher: () => Promise.resolve({ malformed: true }) }),
+    );
+
+    expect(collection.freshness).toBe('unavailable');
+    expect(collection.data).toBeNull();
+    expect(collection.failure).toEqual({ kind: 'invalidated', reason: 'schema-mismatch' });
+    expect(collection.canMutate).toBe(false);
+  });
+
+  it('keeps the previous snapshot as labeled stale when a later payload mismatches', async () => {
+    let call = 0;
+    const collection = await renderCollection(
+      taskOptions({
+        fetcher: () => {
+          call += 1;
+          return call === 1 ? Promise.resolve(taskFixtures) : Promise.resolve('garbage');
+        },
+      }),
+    );
+    expect(collection.freshness).toBe('current');
+
+    await act(async () => {
+      await collection.revalidate();
+    });
+
+    expect(latest?.freshness).toBe('stale');
+    expect(latest?.data).toEqual(taskFixtures);
+    expect(latest?.failure).toEqual({ kind: 'invalidated', reason: 'schema-mismatch' });
+    expect(latest?.canMutate).toBe(false);
+  });
+
+  it('drops the snapshot when the workspace changes under it (cross-workspace)', async () => {
+    let call = 0;
+    const collection = await renderCollection(
+      taskOptions({
+        fetcher: () => {
+          call += 1;
+          return Promise.resolve(
+            call === 1 ? projectFixtures : [{ ...projectFixtures[0], userId: 'user-2' }],
+          );
+        },
+        validate: validateProjectCollection as unknown as (value: unknown) => {
+          data: Task[];
+          workspace: string | null;
+        },
+        source: 'gateway:/api/projects',
+      }),
+    );
+    expect(collection.freshness).toBe('current');
+
+    await act(async () => {
+      await collection.revalidate();
+    });
+
+    expect(latest?.freshness).toBe('unavailable');
+    expect(latest?.data).toBeNull();
+    expect(latest?.failure).toEqual({ kind: 'invalidated', reason: 'cross-workspace' });
+  });
+
+  it('ages from current to stale and refuses mutations on stale data', async () => {
+    let fakeNow = NOW;
+    const collection = await renderCollection(
+      taskOptions({
+        clock: () => fakeNow,
+        policy: { staleAfterMs: 40 },
+        tickMs: 10,
+      }),
+    );
+    expect(collection.freshness).toBe('current');
+
+    // Age the snapshot past the policy and let the tick recompute.
+    fakeNow = NOW + 60;
+    await act(async () => {
+      await new Promise((resolve) => setTimeout(resolve, 25));
+    });
+
+    expect(latest?.freshness).toBe('stale');
+    expect(latest?.data).toEqual(taskFixtures);
+    expect(latest?.canMutate).toBe(false);
+
+    const operation = vi.fn(async () => 'result');
+    await expect(latest?.mutate(operation)).rejects.toBeInstanceOf(StaleMutationError);
+    expect(operation).not.toHaveBeenCalled();
+  });
+
+  it('recovers to current after a successful revalidation', async () => {
+    let call = 0;
+    const collection = await renderCollection(
+      taskOptions({
+        fetcher: () => {
+          call += 1;
+          return call === 1
+            ? Promise.reject(new Error('first attempt failed'))
+            : Promise.resolve(taskFixtures);
+        },
+      }),
+    );
+    expect(collection.freshness).toBe('unavailable');
+
+    await act(async () => {
+      await collection.revalidate();
+    });
+
+    expect(latest?.freshness).toBe('current');
+    expect(latest?.failure).toBeNull();
+
+    const operation = vi.fn(async (data: Task[]) => data.length);
+    await expect(latest?.mutate(operation)).resolves.toBe(taskFixtures.length);
+    expect(operation).toHaveBeenCalledOnce();
+  });
+
+  it('restores a cached snapshot as unverified stale data, then verifies it', async () => {
+    const seededVersion = seedCache('tasks');
+    const deferred = createDeferred<Task[]>();
+    const collection = await renderCollection(taskOptions({ fetcher: () => deferred.promise }));
+
+    // Restored data is situational awareness only: labeled stale, never
+    // current, and mutations are refused before verification.
+    expect(collection.freshness).toBe('stale');
+    expect(collection.data).toEqual(taskFixtures);
+    expect(collection.canMutate).toBe(false);
+    await expect(collection.mutate(vi.fn())).rejects.toBeInstanceOf(StaleMutationError);
+
+    await act(async () => {
+      deferred.resolve(taskFixtures);
+      await deferred.promise;
+    });
+
+    expect(latest?.freshness).toBe('current');
+    expect(latest?.snapshot?.version).toBe(seededVersion + 1);
+  });
+
+  it('never promotes corrupted cache data to current (cache corruption)', async () => {
+    sessionStorage.setItem('mosaic:freshness:v1:tasks', '{"data":');
+    const collection = await renderCollection(
+      taskOptions({ fetcher: () => Promise.reject(new Error('still down')) }),
+    );
+
+    expect(collection.freshness).toBe('unavailable');
+    expect(collection.data).toBeNull();
+    expect(collection.canMutate).toBe(false);
+    // The corrupted entry is dropped so it cannot come back.
+    expect(sessionStorage.getItem('mosaic:freshness:v1:tasks')).toBeNull();
+  });
+
+  it('refuses mutations while unknown or unavailable — the call itself, not just the button', async () => {
+    const deferred = createDeferred<Task[]>();
+    const unknown = await renderCollection(taskOptions({ fetcher: () => deferred.promise }));
+    const operation = vi.fn(async () => 'result');
+    await expect(unknown.mutate(operation)).rejects.toBeInstanceOf(StaleMutationError);
+    expect(operation).not.toHaveBeenCalled();
+    await act(async () => {
+      deferred.reject(new Error('failed'));
+      await deferred.promise.catch(() => undefined);
+    });
+
+    const unavailable = latest!;
+    await expect(unavailable.mutate(operation)).rejects.toBeInstanceOf(StaleMutationError);
+    expect(operation).not.toHaveBeenCalled();
+    expect(unavailable.canMutate).toBe(false);
+  });
+
+  it('degrades to stale with last-known data when a revalidation fails after success', async () => {
+    let call = 0;
+    const collection = await renderCollection(
+      taskOptions({
+        fetcher: () => {
+          call += 1;
+          return call === 1
+            ? Promise.resolve(taskFixtures)
+            : Promise.reject(new Error('connection lost'));
+        },
+      }),
+    );
+    expect(collection.freshness).toBe('current');
+
+    await act(async () => {
+      await collection.revalidate();
+    });
+
+    expect(latest?.freshness).toBe('stale');
+    expect(latest?.data).toEqual(taskFixtures);
+    const failure: FreshnessFailure | null = latest?.failure ?? null;
+    expect(failure).toEqual({ kind: 'fetch', message: 'connection lost' });
+  });
+});

apps/web/src/lib/freshness/use-fresh-collection.ts

@@ -0,0 +1,281 @@
+import { useCallback, useEffect, useMemo, useRef, useState } from 'react';
+import {
+  acceptSnapshot,
+  assertMutable,
+  computeFreshness,
+  DEFAULT_FRESHNESS_POLICY,
+  invalidationReasonLabels,
+  type FreshPayload,
+  type FreshSnapshot,
+  type FreshnessPolicy,
+  type FreshnessState,
+  type InvalidationReason,
+  StaleMutationError,
+} from './model';
+import { clearSnapshotCache, readSnapshotCache, writeSnapshotCache } from './snapshot-cache';
+
+/**
+ * Freshness-aware collection fetch hook (RI-5-001).
+ *
+ * One hook owns one gateway collection end to end: fetch, schema validation,
+ * snapshot acceptance with provenance, session-scoped last-known caching,
+ * aging, and the mutation guard. Pages consume `freshness` and never infer
+ * health from emptiness.
+ */
+
+/** Why the latest validation did not produce a current snapshot. */
+export type FreshnessFailure =
+  | { readonly kind: 'fetch'; readonly message: string }
+  | { readonly kind: 'invalidated'; readonly reason: InvalidationReason };
+
+export interface UseFreshCollectionOptions<T> {
+  /** Source identity for provenance labels, e.g. `gateway:/api/tasks`. */
+  readonly source: string;
+  /** Performs the unvalidated fetch. The hook owns abort and verification. */
+  readonly fetcher: (signal: AbortSignal) => Promise<unknown>;
+  /**
+   * Runtime schema validator. Returning `null` invalidates the payload
+   * (`schema-mismatch`) instead of letting malformed JSON flow into render.
+   */
+  readonly validate: (value: unknown) => FreshPayload<T> | null;
+  /** Overrides of the default freshness policy. */
+  readonly policy?: Partial<FreshnessPolicy>;
+  /**
+   * Session cache key for last-known snapshots. `null`/omitted disables
+   * restore. Restored snapshots are unverified: they render only as
+   * labeled `stale` data until a fetch re-verifies them.
+   */
+  readonly cacheKey?: string | null;
+  /** Injectable clock for deterministic age transitions in tests. */
+  readonly clock?: () => number;
+  /** Aging tick interval override (default derived from `staleAfterMs`). */
+  readonly tickMs?: number;
+  /** When false, no fetch runs (surfaces stay `unavailable`/`unknown`). */
+  readonly enabled?: boolean;
+}
+
+export interface FreshCollection<T> {
+  /** Last verified (or restored-unverified) snapshot, or `null`. */
+  readonly snapshot: FreshSnapshot<T> | null;
+  /** Snapshot data or `null` — never a fabricated empty collection. */
+  readonly data: T | null;
+  readonly freshness: FreshnessState;
+  /** True while a validation request is in flight. */
+  readonly validating: boolean;
+  /** Outcome of the latest failed validation, `null` when healthy. */
+  readonly failure: FreshnessFailure | null;
+  /** False unless freshness is `current`; drives disabled UI affordances. */
+  readonly canMutate: boolean;
+  /** Re-run the fetch and re-verify. Always allowed (it is a read). */
+  readonly revalidate: () => Promise<void>;
+  /**
+   * Run a state-changing operation against verified-current data only.
+   * Rejects with `StaleMutationError` on any other state — the guard fires
+   * even if a disabled button was bypassed (defense in depth).
+   */
+  readonly mutate: <R>(operation: (data: T) => Promise<R>) => Promise<R>;
+}
+
+const defaultClock = (): number => Date.now();
+
+function resolveTickMs(policy: FreshnessPolicy, override?: number): number {
+  if (override !== undefined && override > 0) return override;
+  return Math.min(5_000, Math.max(250, Math.floor(policy.staleAfterMs / 4)));
+}
+
+function isAuthFailure(caught: unknown): boolean {
+  return (
+    typeof caught === 'object' &&
+    caught !== null &&
+    'statusCode' in caught &&
+    ((caught as { statusCode?: unknown }).statusCode === 401 ||
+      (caught as { statusCode?: unknown }).statusCode === 403)
+  );
+}
+
+function fetchFailureMessage(caught: unknown): string {
+  if (caught instanceof Error && caught.message.trim().length > 0) return caught.message;
+  return 'The request failed.';
+}
+
+/** Human-readable summary of a failure for unavailable/stale notices. */
+export function describeFailure(failure: FreshnessFailure | null): string | null {
+  if (failure === null) return null;
+  if (failure.kind === 'fetch') return failure.message;
+  return `The snapshot was invalidated: ${invalidationReasonLabels[failure.reason]}.`;
+}
+
+export function useFreshCollection<T>(options: UseFreshCollectionOptions<T>): FreshCollection<T> {
+  const optionsRef = useRef(options);
+  optionsRef.current = options;
+
+  const policy = useMemo<FreshnessPolicy>(
+    () => ({ ...DEFAULT_FRESHNESS_POLICY, ...options.policy }),
+    [options.policy],
+  );
+  const policyRef = useRef(policy);
+  policyRef.current = policy;
+
+  const clockRef = useRef(options.clock ?? defaultClock);
+  clockRef.current = options.clock ?? defaultClock;
+
+  const [snapshot, setSnapshot] = useState<FreshSnapshot<T> | null>(null);
+  const [failure, setFailure] = useState<FreshnessFailure | null>(null);
+  const [unverified, setUnverified] = useState(false);
+  const [validating, setValidating] = useState(options.enabled !== false);
+  const [now, setNow] = useState(() => (options.clock ?? defaultClock)());
+
+  const snapshotRef = useRef(snapshot);
+  snapshotRef.current = snapshot;
+  const failureRef = useRef(failure);
+  failureRef.current = failure;
+  const unverifiedRef = useRef(unverified);
+  unverifiedRef.current = unverified;
+
+  const runRef = useRef(0);
+  const abortRef = useRef<AbortController | null>(null);
+
+  const revalidate = useCallback(async (): Promise<void> => {
+    const current = optionsRef.current;
+    if (current.enabled === false) {
+      setValidating(false);
+      return;
+    }
+
+    const runId = ++runRef.current;
+    abortRef.current?.abort();
+    const controller = new AbortController();
+    abortRef.current = controller;
+    setValidating(true);
+
+    let value: unknown;
+    try {
+      value = await current.fetcher(controller.signal);
+    } catch (caught) {
+      if (runRef.current !== runId || controller.signal.aborted) return;
+      if (isAuthFailure(caught)) {
+        // An unauthenticated viewer must not keep (or be served) the
+        // previous user's last-known data.
+        setSnapshot(null);
+        setUnverified(false);
+        if (current.cacheKey) clearSnapshotCache(current.cacheKey);
+      }
+      setFailure({ kind: 'fetch', message: fetchFailureMessage(caught) });
+      setValidating(false);
+      return;
+    }
+
+    if (runRef.current !== runId) return;
+
+    const result = acceptSnapshot({
+      value,
+      validate: current.validate,
+      previous: snapshotRef.current,
+      policy: policyRef.current,
+      source: current.source,
+      now: clockRef.current(),
+    });
+
+    if (result.outcome === 'accepted') {
+      setSnapshot(result.snapshot);
+      setUnverified(false);
+      setFailure(null);
+      if (current.cacheKey) writeSnapshotCache(current.cacheKey, result.snapshot);
+    } else {
+      if (result.reason === 'cross-workspace') {
+        // Data verified for a different workspace must not linger as
+        // last-known situational awareness either.
+        setSnapshot(null);
+        setUnverified(false);
+      }
+      if (current.cacheKey) clearSnapshotCache(current.cacheKey);
+      setFailure({ kind: 'invalidated', reason: result.reason });
+    }
+    setValidating(false);
+  }, []);
+
+  // Restore the last-known snapshot (unverified) and run the first fetch.
+  useEffect(() => {
+    if (optionsRef.current.enabled === false) {
+      setValidating(false);
+      return;
+    }
+
+    const cacheKey = optionsRef.current.cacheKey;
+    if (cacheKey) {
+      const restored = readSnapshotCache<T>({
+        key: cacheKey,
+        workspace: policyRef.current.workspace,
+        policy: policyRef.current,
+        validate: optionsRef.current.validate,
+      });
+      if (restored.outcome === 'hit') {
+        setSnapshot(restored.snapshot);
+        setUnverified(true);
+      } else if (restored.outcome === 'invalidated') {
+        // A corrupted/foreign/regressed entry is dropped immediately; it must
+        // never surface as data. The fetch decides the visible state.
+        clearSnapshotCache(cacheKey);
+      }
+    }
+
+    void revalidate();
+
+    return () => {
+      abortRef.current?.abort();
+    };
+    // Mount-once by design: `revalidate` is stable and reads live options
+    // through refs, so it never needs to re-run when options change.
+    // Route-param pages remount this hook via an identity `key` instead.
+  }, [revalidate]);
+
+  // Aging tick: recomputes freshness as the snapshot ages past the policy.
+  useEffect(() => {
+    const interval = setInterval(
+      () => {
+        setNow(clockRef.current());
+      },
+      resolveTickMs(policyRef.current, optionsRef.current.tickMs),
+    );
+    return () => clearInterval(interval);
+  }, []);
+
+  const freshness = useMemo<FreshnessState>(() => {
+    if (snapshot === null) return validating ? 'unknown' : 'unavailable';
+    return computeFreshness({
+      snapshot,
+      policy,
+      now,
+      degraded: failure !== null || unverified,
+    });
+    // `now` from state covers age; refs inside computeFreshness are pure.
+  }, [snapshot, validating, failure, unverified, now, policy]);
+
+  const canMutate = freshness === 'current';
+
+  const mutate = useCallback(async <R>(operation: (data: T) => Promise<R>): Promise<R> => {
+    const currentSnapshot = snapshotRef.current;
+    // No verified snapshot at all: with nothing verified there is nothing
+    // current to mutate, regardless of the recorded failure.
+    if (currentSnapshot === null) throw new StaleMutationError('unavailable');
+    const state = computeFreshness({
+      snapshot: currentSnapshot,
+      policy: policyRef.current,
+      now: clockRef.current(),
+      degraded: failureRef.current !== null || unverifiedRef.current,
+    });
+    assertMutable(state);
+    return operation(currentSnapshot.data);
+  }, []);
+
+  return {
+    snapshot,
+    data: snapshot === null ? null : snapshot.data,
+    freshness,
+    validating,
+    failure,
+    canMutate,
+    revalidate,
+    mutate,
+  };
+}

apps/web/src/lib/freshness/validators.spec.ts

@@ -0,0 +1,103 @@
+import { describe, expect, it } from 'vitest';
+import type { Mission, Project, Task } from '@/lib/types';
+import {
+  validateMissionCollection,
+  validateProjectCollection,
+  validateProjectEntity,
+  validateTaskCollection,
+} from './validators';
+import { missionFixtures, projectFixtures, taskFixtures } from '@/spa/pages/page-fixtures';
+
+describe('validateTaskCollection', () => {
+  it('accepts a well-formed task collection', () => {
+    expect(validateTaskCollection(taskFixtures)).toEqual({
+      data: taskFixtures,
+      workspace: null,
+    });
+  });
+
+  it('accepts an empty collection (a healthy empty state is a valid payload)', () => {
+    expect(validateTaskCollection([])).toEqual({ data: [], workspace: null });
+  });
+
+  it.each([
+    ['not an array', { items: [] }],
+    ['item is not an object', ['nope']],
+    ['missing id', [{ ...(taskFixtures[0] as Task), id: undefined }]],
+    ['missing title', [{ ...(taskFixtures[0] as Task), title: undefined }]],
+    ['unknown status enum', [{ ...(taskFixtures[0] as Task), status: 'finished' }]],
+    ['unknown priority enum', [{ ...(taskFixtures[0] as Task), priority: 'urgent' }]],
+    ['tags of the wrong type', [{ ...(taskFixtures[0] as Task), tags: 'spa' }]],
+    ['metadata of the wrong type', [{ ...(taskFixtures[0] as Task), metadata: 'notes' }]],
+    ['createdAt of the wrong type', [{ ...(taskFixtures[0] as Task), createdAt: 1234 }]],
+    ['null sneaks past a required string', [{ ...(taskFixtures[0] as Task), title: null }]],
+  ])('rejects a malformed payload: %s', (_label, value) => {
+    expect(validateTaskCollection(value)).toBeNull();
+  });
+});
+
+describe('validateMissionCollection', () => {
+  it('accepts a well-formed mission collection', () => {
+    expect(validateMissionCollection(missionFixtures)).toEqual({
+      data: missionFixtures,
+      workspace: null,
+    });
+  });
+
+  it.each([
+    ['not an array', null],
+    ['item missing name', [{ ...(missionFixtures[0] as Mission), name: 42 }]],
+    ['unknown status enum', [{ ...(missionFixtures[0] as Mission), status: 'canceled' }]],
+    ['projectId of the wrong type', [{ ...(missionFixtures[0] as Mission), projectId: 7 }]],
+  ])('rejects a malformed payload: %s', (_label, value) => {
+    expect(validateMissionCollection(value)).toBeNull();
+  });
+});
+
+describe('validateProjectCollection', () => {
+  it('accepts a uniform workspace-scoped collection and reports its workspace', () => {
+    expect(validateProjectCollection(projectFixtures)).toEqual({
+      data: projectFixtures,
+      workspace: 'user-1',
+    });
+  });
+
+  it('accepts an empty collection with no workspace identity', () => {
+    expect(validateProjectCollection([])).toEqual({ data: [], workspace: null });
+  });
+
+  it.each([
+    ['not an array', 42],
+    ['item missing userId', [{ ...(projectFixtures[0] as Project), userId: undefined }]],
+    ['unknown status enum', [{ ...(projectFixtures[0] as Project), status: 'live' }]],
+    ['description of the wrong type', [{ ...(projectFixtures[0] as Project), description: 1 }]],
+  ])('rejects a malformed payload: %s', (_label, value) => {
+    expect(validateProjectCollection(value)).toBeNull();
+  });
+
+  it('rejects a collection mixing workspace identities (cross-workspace leak)', () => {
+    const mixed = [
+      projectFixtures[0] as Project,
+      { ...(projectFixtures[1] as Project), userId: 'user-2' },
+    ];
+    expect(validateProjectCollection(mixed)).toBeNull();
+  });
+});
+
+describe('validateProjectEntity', () => {
+  it('accepts a well-formed project and reports its workspace', () => {
+    expect(validateProjectEntity(projectFixtures[0])).toEqual({
+      data: projectFixtures[0],
+      workspace: 'user-1',
+    });
+  });
+
+  it.each([
+    ['not an object', 'project-1'],
+    ['null', null],
+    ['array', [projectFixtures[0]]],
+    ['missing userId', [{ ...(projectFixtures[0] as Project), userId: null }]],
+  ])('rejects a malformed entity: %s', (_label, value) => {
+    expect(validateProjectEntity(value)).toBeNull();
+  });
+});

apps/web/src/lib/freshness/validators.ts

@@ -0,0 +1,135 @@
+import type { Mission, Project, Task, MissionStatus, TaskPriority, TaskStatus } from '@/lib/types';
+import type { FreshPayload } from './model';
+
+/**
+ * Runtime schema validators for gateway collections (RI-5-001).
+ *
+ * `api<T>()` returns untrusted JSON cast to `T`; these validators are the
+ * seam where a malformed response becomes an explicit schema mismatch
+ * instead of flowing into the render path as if it were healthy data.
+ */
+
+const taskStatuses: readonly TaskStatus[] = [
+  'not-started',
+  'in-progress',
+  'blocked',
+  'done',
+  'cancelled',
+];
+const taskPriorities: readonly TaskPriority[] = ['critical', 'high', 'medium', 'low'];
+const missionStatuses: readonly MissionStatus[] = [
+  'planning',
+  'active',
+  'paused',
+  'completed',
+  'failed',
+];
+const projectStatuses: readonly Project['status'][] = ['active', 'paused', 'completed', 'archived'];
+
+function isRecord(value: unknown): value is Record<string, unknown> {
+  return typeof value === 'object' && value !== null && !Array.isArray(value);
+}
+
+function isString(value: unknown): value is string {
+  return typeof value === 'string';
+}
+
+function isNullableString(value: unknown): value is string | null {
+  return value === null || typeof value === 'string';
+}
+
+function isOneOf<T extends string>(value: unknown, allowed: readonly T[]): value is T {
+  return typeof value === 'string' && (allowed as readonly string[]).includes(value);
+}
+
+function isNullableRecord(value: unknown): value is Record<string, unknown> | null {
+  return value === null || isRecord(value);
+}
+
+function isNullableStringArray(value: unknown): value is string[] | null {
+  if (value === null) return true;
+  if (!Array.isArray(value)) return false;
+  return value.every((item) => typeof item === 'string');
+}
+
+function isIsoLike(value: unknown): value is string {
+  return typeof value === 'string' && value.length > 0;
+}
+
+function isTask(value: unknown): value is Task {
+  if (!isRecord(value)) return false;
+  return (
+    isString(value['id']) &&
+    isString(value['title']) &&
+    isOneOf(value['status'], taskStatuses) &&
+    isOneOf(value['priority'], taskPriorities) &&
+    isNullableString(value['projectId']) &&
+    isNullableString(value['missionId']) &&
+    isNullableString(value['assignee']) &&
+    isNullableStringArray(value['tags']) &&
+    isNullableRecord(value['metadata']) &&
+    isNullableString(value['dueDate']) &&
+    isIsoLike(value['createdAt']) &&
+    isIsoLike(value['updatedAt'])
+  );
+}
+
+/** Tasks carry no workspace identity; scope falls back to the policy. */
+export function validateTaskCollection(value: unknown): FreshPayload<Task[]> | null {
+  if (!Array.isArray(value) || !value.every(isTask)) return null;
+  return { data: value as Task[], workspace: null };
+}
+
+function isMission(value: unknown): value is Mission {
+  if (!isRecord(value)) return false;
+  return (
+    isString(value['id']) &&
+    isString(value['name']) &&
+    isOneOf(value['status'], missionStatuses) &&
+    isNullableString(value['projectId']) &&
+    isNullableString(value['description']) &&
+    isNullableRecord(value['metadata']) &&
+    isIsoLike(value['createdAt']) &&
+    isIsoLike(value['updatedAt'])
+  );
+}
+
+/** Missions carry no workspace identity; scope falls back to the policy. */
+export function validateMissionCollection(value: unknown): FreshPayload<Mission[]> | null {
+  if (!Array.isArray(value) || !value.every(isMission)) return null;
+  return { data: value as Mission[], workspace: null };
+}
+
+function isProject(value: unknown): value is Project {
+  if (!isRecord(value)) return false;
+  return (
+    isString(value['id']) &&
+    isString(value['name']) &&
+    isOneOf(value['status'], projectStatuses) &&
+    isString(value['userId']) &&
+    isNullableString(value['description']) &&
+    isNullableRecord(value['metadata']) &&
+    isIsoLike(value['createdAt']) &&
+    isIsoLike(value['updatedAt'])
+  );
+}
+
+/**
+ * Projects are workspace-scoped: every item must carry the same `userId`.
+ * A collection mixing identities (cross-workspace leak) is a schema
+ * mismatch; the uniform `userId` becomes the snapshot workspace.
+ */
+export function validateProjectCollection(value: unknown): FreshPayload<Project[]> | null {
+  if (!Array.isArray(value) || !value.every(isProject)) return null;
+  const projects = value as Project[];
+  const workspaces = new Set(projects.map((project) => project.userId));
+  if (workspaces.size > 1) return null;
+  return { data: projects, workspace: projects.length > 0 ? projects[0]!.userId : null };
+}
+
+/** Single project entity (project detail primary collection). */
+export function validateProjectEntity(value: unknown): FreshPayload<Project> | null {
+  if (!isProject(value)) return null;
+  const project = value as Project;
+  return { data: project, workspace: project.userId };
+}

apps/web/src/spa/pages/project-detail.spec.tsx

@@ -35,6 +35,7 @@ afterEach(async () => {
   document.body.replaceChildren();
   root = null;
   apiMock.mockReset();
+  sessionStorage.clear();
 });
 
 async function renderProjectDetailPage(): Promise<ReturnType<typeof createMemoryRouter>> {
@@ -64,21 +65,49 @@ function clickButtonByText(text: string): void {
   button.dispatchEvent(new MouseEvent('click', { bubbles: true }));
 }
 
+async function flushAct(): Promise<void> {
+  await act(async () => {
+    await Promise.resolve();
+  });
+}
+
+interface Deferred<T> {
+  promise: Promise<T>;
+  resolve: (value: T) => void;
+}
+
+function createDeferred<T>(): Deferred<T> {
+  let resolve!: (value: T) => void;
+  const promise = new Promise<T>((res) => {
+    resolve = res;
+  });
+  return { promise, resolve };
+}
+
+const projectOneTasks = taskFixtures.filter((task) => task.projectId === 'project-1');
+
+function mockHealthyLoad(): void {
+  apiMock
+    .mockResolvedValueOnce(projectFixtures[0])
+    .mockResolvedValueOnce(missionFixtures)
+    .mockResolvedValueOnce(projectOneTasks);
+}
+
 describe('ProjectDetailPage', () => {
   it('loads the project, tasks, missions, and optional PRD content for the active project', async () => {
-    apiMock
-      .mockResolvedValueOnce(projectFixtures[0])
-      .mockResolvedValueOnce(missionFixtures)
-      .mockResolvedValueOnce(taskFixtures.filter((task) => task.projectId === 'project-1'));
+    mockHealthyLoad();
 
     await renderProjectDetailPage();
 
-    expect(apiMock.mock.calls).toEqual([
-      ['/api/projects/project-1'],
-      ['/api/missions'],
-      ['/api/tasks?projectId=project-1'],
+    expect(apiMock.mock.calls.map((call) => call[0])).toEqual([
+      '/api/projects/project-1',
+      '/api/missions',
+      '/api/tasks?projectId=project-1',
     ]);
 
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
     expect(container.textContent).toContain('Mosaic Stack');
     expect(container.textContent).toContain('Route /projects/:id');
     expect(container.textContent).toContain('Tasks');
@@ -101,10 +130,7 @@ describe('ProjectDetailPage', () => {
   });
 
   it('opens and closes the existing read-only task modal from the tasks tab', async () => {
-    apiMock
-      .mockResolvedValueOnce(projectFixtures[0])
-      .mockResolvedValueOnce(missionFixtures)
-      .mockResolvedValueOnce(taskFixtures.filter((task) => task.projectId === 'project-1'));
+    mockHealthyLoad();
 
     await renderProjectDetailPage();
 
@@ -134,35 +160,153 @@ describe('ProjectDetailPage', () => {
     expect(container.querySelector('[role="dialog"]')).toBeNull();
   });
 
-  it('renders the project with an empty missions tab when the missions request fails', async () => {
-    apiMock
-      .mockResolvedValueOnce(projectFixtures[0])
-      .mockRejectedValueOnce(new Error('Missions request failed'))
-      .mockResolvedValueOnce(taskFixtures.filter((task) => task.projectId === 'project-1'));
+  it('shows verified completion verdicts when the task collection is current', async () => {
+    mockHealthyLoad();
 
     await renderProjectDetailPage();
 
-    expect(container.textContent).toContain('Mosaic Stack');
-    expect(container.querySelector('[role="alert"]')).toBeNull();
-
-    await act(async () => {
-      clickButtonByText('Missions (0)');
-    });
-
-    expect(container.textContent).toContain('No missions for this project');
+    const doneCard = [...container.querySelectorAll('div')].find(
+      (candidate) => candidate.textContent === 'Done1',
+    );
+    expect(doneCard).toBeTruthy();
+    const inProgressCard = [...container.querySelectorAll('div')].find(
+      (candidate) => candidate.textContent === 'In Progress1',
+    );
+    expect(inProgressCard).toBeTruthy();
   });
 
-  it('renders a visible alert when the project request fails and lets the user navigate back', async () => {
+  it('renders an explicit unavailable missions tab when the missions request fails (partial, not empty)', async () => {
+    apiMock
+      .mockResolvedValueOnce(projectFixtures[0])
+      .mockRejectedValueOnce(new Error('Missions request failed'))
+      .mockResolvedValueOnce(projectOneTasks);
+
+    await renderProjectDetailPage();
+
+    // Secondary failure degrades the surface to partial; the project itself
+    // still renders.
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'partial',
+    );
+    expect(container.textContent).toContain('Mosaic Stack');
+    const partial = container.querySelector('[role="status"]');
+    expect(partial?.textContent).toContain('Missions');
+    expect(partial?.textContent).toContain('unavailable');
+
+    await act(async () => {
+      clickButtonByText('Missions (?)');
+    });
+
+    const alert = container.querySelector('[role="alert"]');
+    expect(alert?.textContent).toContain('Missions request failed');
+    // Negative control: a failed fetch must not look like an empty list.
+    expect(container.textContent).not.toContain('No missions for this project');
+  });
+
+  it('marks derived verdicts unknown when the tasks collection is unavailable', async () => {
+    apiMock
+      .mockResolvedValueOnce(projectFixtures[0])
+      .mockResolvedValueOnce(missionFixtures)
+      .mockRejectedValueOnce(new Error('Tasks request failed'));
+
+    await renderProjectDetailPage();
+
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'partial',
+    );
+
+    // Completion verdicts become unknown ('?') — never green counts.
+    for (const label of ['Done', 'In Progress', 'Blocked', 'Tasks']) {
+      const unknownCard = [...container.querySelectorAll('div')].find(
+        (candidate) => candidate.textContent === `${label}?`,
+      );
+      expect(unknownCard, `expected ${label} card to render ?`).toBeTruthy();
+    }
+    // Negative control: no green "Done 1" verdict anywhere.
+    expect(
+      [...container.querySelectorAll('div')].some((candidate) => candidate.textContent === 'Done1'),
+    ).toBe(false);
+
+    await act(async () => {
+      clickButtonByText('Tasks (?)');
+    });
+
+    const alert = container.querySelector('[role="alert"]');
+    expect(alert?.textContent).toContain('Tasks request failed');
+    // Negative control: no healthy empty task list from a failed fetch.
+    expect(container.textContent).not.toContain('No tasks found');
+    expect(container.querySelector('table')).toBeNull();
+  });
+
+  it('recovers a partial surface to current after revalidation', async () => {
+    apiMock
+      .mockResolvedValueOnce(projectFixtures[0])
+      .mockResolvedValueOnce(missionFixtures)
+      .mockRejectedValueOnce(new Error('Tasks request failed'))
+      .mockResolvedValueOnce(projectFixtures[0])
+      .mockResolvedValueOnce(missionFixtures)
+      .mockResolvedValueOnce(projectOneTasks);
+
+    await renderProjectDetailPage();
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'partial',
+    );
+
+    await act(async () => {
+      clickButtonByText('Revalidate');
+    });
+    await flushAct();
+
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
+    expect(
+      [...container.querySelectorAll('div')].some((candidate) => candidate.textContent === 'Done1'),
+    ).toBe(true);
+  });
+
+  it("never shows one project's data on another project's route after navigation", async () => {
+    mockHealthyLoad();
+
+    const router = await renderProjectDetailPage();
+    expect(container.textContent).toContain('Mosaic Stack');
+
+    const deferred = createDeferred<(typeof projectFixtures)[number]>();
+    apiMock
+      .mockResolvedValueOnce(deferred.promise)
+      .mockResolvedValueOnce([])
+      .mockResolvedValueOnce([]);
+
+    await act(async () => {
+      await router.navigate('/projects/project-2');
+    });
+
+    // While project-2 loads, nothing from project-1 may render on its route.
+    expect(container.textContent).toContain('Loading project...');
+    expect(container.textContent).not.toContain('Mosaic Stack');
+    expect(container.textContent).not.toContain('Route /projects/:id');
+
+    await act(async () => {
+      deferred.resolve(projectFixtures[1]!);
+      await deferred.promise;
+    });
+
+    expect(container.textContent).toContain('Agent Runtime');
+    expect(apiMock.mock.calls[3]?.[0]).toBe('/api/projects/project-2');
+  });
+
+  it('renders a visible unavailable state when the project request fails and lets the user navigate back', async () => {
     apiMock
       .mockRejectedValueOnce(new Error('Project request failed'))
       .mockResolvedValueOnce(missionFixtures)
-      .mockResolvedValueOnce(taskFixtures.filter((task) => task.projectId === 'project-1'));
+      .mockResolvedValueOnce(projectOneTasks);
 
     const router = await renderProjectDetailPage();
 
     const alert = container.querySelector('[role="alert"]');
     expect(alert).toBeTruthy();
     expect(alert?.textContent).toContain('Project request failed');
+    expect(alert?.textContent).toContain('not an empty result');
     expect(container.textContent).not.toContain('Mosaic Stack');
 
     await act(async () => {

apps/web/src/spa/pages/project-detail.tsx

@@ -1,14 +1,30 @@
-import { useEffect, useState, type ReactElement } from 'react';
+import { useState, type ReactElement } from 'react';
 import { useNavigate, useParams } from 'react-router-dom';
 import { MissionTimeline } from '@/components/projects/mission-timeline';
 import { PrdViewer } from '@/components/projects/prd-viewer';
 import { TaskDetailModal } from '@/components/tasks/task-detail-modal';
 import { TaskListView } from '@/components/tasks/task-list-view';
 import { TaskStatusSummary } from '@/components/tasks/task-status-summary';
+import {
+  PartialDataNotice,
+  StaleDataNotice,
+  UnavailableDataNotice,
+} from '@/components/freshness/freshness-notices';
 import { api } from '@/lib/api';
 import { cn } from '@/lib/cn';
 import type { Mission, Project, Task, TaskStatus } from '@/lib/types';
-import { getErrorMessage } from './page-errors';
+import {
+  combineFreshness,
+  UNKNOWN_VERDICT,
+  verdictValue,
+  type FreshSnapshot,
+} from '@/lib/freshness/model';
+import { describeFailure, useFreshCollection } from '@/lib/freshness/use-fresh-collection';
+import {
+  validateMissionCollection,
+  validateProjectEntity,
+  validateTaskCollection,
+} from '@/lib/freshness/validators';
 
 type Tab = 'overview' | 'tasks' | 'missions' | 'prd';
 
@@ -51,73 +67,62 @@ function TabButton({ id, label, activeTab, onClick }: TabButtonProps): ReactElem
   );
 }
 
+/** Remounts per project id so no state from one project renders for another. */
 export function ProjectDetailPage(): ReactElement {
   const { id = '' } = useParams();
+  return <ProjectDetail id={id} key={id} />;
+}
+
+function ProjectDetail({ id }: { id: string }): ReactElement {
   const navigate = useNavigate();
-  const [project, setProject] = useState<Project | null>(null);
-  const [missions, setMissions] = useState<Mission[]>([]);
-  const [tasks, setTasks] = useState<Task[]>([]);
-  const [loading, setLoading] = useState(true);
-  const [error, setError] = useState<string | null>(null);
+  const enabled = id.length > 0;
+
+  // Primary collection gates the surface; missions and tasks are secondaries
+  // whose failures degrade the surface to `partial` instead of rendering
+  // empty healthy lists.
+  const project = useFreshCollection<Project>({
+    source: `gateway:/api/projects/${id}`,
+    fetcher: (signal) => api<unknown>(`/api/projects/${id}`, { signal }),
+    validate: validateProjectEntity,
+    // No last-known restore: the entity carries workspace identity that
+    // cannot be scope-checked before display (see ProjectsPage note).
+    enabled,
+  });
+  const missions = useFreshCollection<Mission[]>({
+    source: 'gateway:/api/missions',
+    fetcher: (signal) => api<unknown>('/api/missions', { signal }),
+    validate: validateMissionCollection,
+    cacheKey: enabled ? 'missions' : null,
+    enabled,
+  });
+  const tasks = useFreshCollection<Task[]>({
+    source: `gateway:/api/tasks?projectId=${id}`,
+    fetcher: (signal) => api<unknown>(`/api/tasks?projectId=${id}`, { signal }),
+    validate: validateTaskCollection,
+    cacheKey: enabled ? `project-tasks:${id}` : null,
+    enabled,
+  });
+
   const [activeTab, setActiveTab] = useState<Tab>('overview');
   const [taskFilter, setTaskFilter] = useState<TaskStatus | 'all'>('all');
   const [selectedTask, setSelectedTask] = useState<Task | null>(null);
 
-  useEffect(() => {
-    if (!id) {
-      setError('Project id is missing.');
-      setLoading(false);
-      return;
-    }
+  const surface = combineFreshness(project.freshness, [missions.freshness, tasks.freshness]);
+  const tasksVerified = tasks.freshness === 'current';
+  const projectMissions = missions.data?.filter((mission) => mission.projectId === id) ?? null;
 
-    let cancelled = false;
-    setLoading(true);
-    setError(null);
+  const retryAll = (): void => {
+    void Promise.all([project.revalidate(), missions.revalidate(), tasks.revalidate()]);
+  };
 
-    void Promise.all([
-      api<Project>('/api/projects/' + id),
-      api<Mission[]>('/api/missions').catch(() => [] as Mission[]),
-      api<Task[]>('/api/tasks?projectId=' + id).catch(() => [] as Task[]),
-    ])
-      .then(([loadedProject, allMissions, loadedTasks]) => {
-        if (cancelled) return;
-        setProject(loadedProject);
-        setMissions(allMissions.filter((mission) => mission.projectId === id));
-        setTasks(loadedTasks);
-      })
-      .catch((caught: unknown) => {
-        if (cancelled) return;
-        setError(getErrorMessage(caught, 'Failed to load project.'));
-      })
-      .finally(() => {
-        if (cancelled) return;
-        setLoading(false);
-      });
-
-    return () => {
-      cancelled = true;
-    };
-  }, [id]);
-
-  if (loading) {
-    return (
-      <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
-        <header className="mb-6 border-b px-1 pb-3">
-          <h1 className="text-2xl font-semibold">Project</h1>
-        </header>
-        <p className="py-16 text-center text-sm text-text-muted">Loading project...</p>
-      </div>
-    );
-  }
-
-  if (error || !project) {
+  if (!enabled) {
     return (
       <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
         <header className="mb-6 border-b px-1 pb-3">
           <h1 className="text-2xl font-semibold">Project</h1>
         </header>
         <div role="alert" className="rounded-lg border border-error/40 px-4 py-3 text-sm">
-          {error ?? 'Project not found.'}
+          Project id is missing.
         </div>
         <button
           type="button"
@@ -130,18 +135,81 @@ export function ProjectDetailPage(): ReactElement {
     );
   }
 
+  if (project.freshness === 'unknown') {
+    return (
+      <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
+        <header className="mb-6 border-b px-1 pb-3">
+          <h1 className="text-2xl font-semibold">Project</h1>
+        </header>
+        <p className="py-16 text-center text-sm text-text-muted">Loading project...</p>
+      </div>
+    );
+  }
+
+  if (project.freshness === 'unavailable' || project.data === null) {
+    return (
+      <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
+        <header className="mb-6 border-b px-1 pb-3">
+          <h1 className="text-2xl font-semibold">Project</h1>
+        </header>
+        <UnavailableDataNotice
+          title="This project"
+          detail={describeFailure(project.failure)}
+          onRetry={retryAll}
+        />
+        <button
+          type="button"
+          onClick={() => navigate('/projects')}
+          className="mt-4 w-fit text-sm underline"
+        >
+          Back to projects
+        </button>
+      </div>
+    );
+  }
+
+  const projectTasks = tasks.data ?? null;
   const filteredTasks =
-    taskFilter === 'all' ? tasks : tasks.filter((task) => task.status === taskFilter);
-  const prdContent = getPrdContent(project);
+    projectTasks === null
+      ? []
+      : taskFilter === 'all'
+        ? projectTasks
+        : projectTasks.filter((task) => task.status === taskFilter);
+
+  // Derived completion verdicts: unknown (never green) unless the task
+  // collection is verified current.
+  const doneCount = projectTasks?.filter((task) => task.status === 'done').length ?? 0;
+  const inProgressCount = projectTasks?.filter((task) => task.status === 'in-progress').length ?? 0;
+  const blockedCount = projectTasks?.filter((task) => task.status === 'blocked').length ?? 0;
+
+  const prdContent = getPrdContent(project.data);
   const tabs: Array<{ id: Tab; label: string }> = [
     { id: 'overview', label: 'Overview' },
-    { id: 'tasks', label: `Tasks (${tasks.length})` },
-    { id: 'missions', label: `Missions (${missions.length})` },
+    {
+      id: 'tasks',
+      label: `Tasks (${projectTasks === null ? UNKNOWN_VERDICT : projectTasks.length})`,
+    },
+    {
+      id: 'missions',
+      label: `Missions (${projectMissions === null ? UNKNOWN_VERDICT : projectMissions.length})`,
+    },
     ...(prdContent ? [{ id: 'prd' as const, label: 'PRD' }] : []),
   ];
 
+  const staleSnapshot: FreshSnapshot<unknown> | null =
+    project.freshness === 'stale'
+      ? project.snapshot
+      : missions.freshness === 'stale'
+        ? missions.snapshot
+        : tasks.freshness === 'stale'
+          ? tasks.snapshot
+          : null;
+  const missingSections: string[] = [];
+  if (missions.freshness === 'unavailable') missingSections.push('Missions');
+  if (tasks.freshness === 'unavailable') missingSections.push('Tasks');
+
   return (
-    <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
+    <div data-freshness={surface} className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
       <header className="mb-6 border-b px-1 pb-3">
         <nav className="mb-4 flex items-center gap-2 text-sm text-text-muted">
           <button
@@ -152,49 +220,64 @@ export function ProjectDetailPage(): ReactElement {
             Projects
           </button>
           <span>/</span>
-          <span className="text-text-primary">{project.name}</span>
+          <span className="text-text-primary">{project.data.name}</span>
         </nav>
 
         <div className="flex items-start justify-between gap-4">
           <div>
             <div className="flex items-center gap-3">
-              <h1 className="text-2xl font-semibold text-text-primary">{project.name}</h1>
+              <h1 className="text-2xl font-semibold text-text-primary">{project.data.name}</h1>
               <span
                 className={cn(
                   'rounded-full px-2 py-0.5 text-xs',
-                  projectStatusColors[project.status] ?? 'bg-gray-600/20 text-gray-400',
+                  projectStatusColors[project.data.status] ?? 'bg-gray-600/20 text-gray-400',
                 )}
               >
-                {project.status}
+                {project.data.status}
               </span>
             </div>
-            {project.description ? (
-              <p className="mt-1 text-sm text-text-muted">{project.description}</p>
+            {project.data.description ? (
+              <p className="mt-1 text-sm text-text-muted">{project.data.description}</p>
             ) : null}
             <p className="mt-2 text-xs text-text-muted">
-              Created {new Date(project.createdAt).toLocaleDateString()} · Updated{' '}
-              {new Date(project.updatedAt).toLocaleDateString()}
+              Created {new Date(project.data.createdAt).toLocaleDateString()} · Updated{' '}
+              {new Date(project.data.updatedAt).toLocaleDateString()}
             </p>
           </div>
         </div>
       </header>
 
+      {staleSnapshot !== null ? (
+        <div className="mb-6">
+          <StaleDataNotice label={staleSnapshot} onRetry={retryAll} />
+        </div>
+      ) : null}
+
+      {missingSections.length > 0 ? (
+        <div className="mb-6">
+          <PartialDataNotice missing={missingSections} onRetry={retryAll} />
+        </div>
+      ) : null}
+
       <div className="mb-6 grid grid-cols-2 gap-3 sm:grid-cols-4">
-        <StatCard label="Tasks" value={String(tasks.length)} />
+        <StatCard
+          label="Tasks"
+          value={projectTasks === null ? UNKNOWN_VERDICT : String(projectTasks.length)}
+        />
         <StatCard
           label="Done"
-          value={String(tasks.filter((task) => task.status === 'done').length)}
-          valueClass="text-success"
+          value={verdictValue(tasksVerified, String(doneCount))}
+          valueClass={tasksVerified ? 'text-success' : undefined}
         />
         <StatCard
           label="In Progress"
-          value={String(tasks.filter((task) => task.status === 'in-progress').length)}
-          valueClass="text-blue-400"
+          value={verdictValue(tasksVerified, String(inProgressCount))}
+          valueClass={tasksVerified ? 'text-blue-400' : undefined}
         />
         <StatCard
           label="Blocked"
-          value={String(tasks.filter((task) => task.status === 'blocked').length)}
-          valueClass={tasks.some((task) => task.status === 'blocked') ? 'text-error' : undefined}
+          value={verdictValue(tasksVerified, String(blockedCount))}
+          valueClass={tasksVerified && blockedCount > 0 ? 'text-error' : undefined}
         />
       </div>
 
@@ -211,23 +294,43 @@ export function ProjectDetailPage(): ReactElement {
       </div>
 
       {activeTab === 'overview' ? (
-        <OverviewTab project={project} missions={missions} tasks={tasks} />
+        <OverviewTab project={project.data} missions={projectMissions} tasks={projectTasks} />
       ) : null}
 
       {activeTab === 'tasks' ? (
         <div>
-          <div className="mb-4">
-            <TaskStatusSummary
-              tasks={tasks}
-              activeFilter={taskFilter}
-              onFilterChange={setTaskFilter}
+          {projectTasks === null ? (
+            <UnavailableDataNotice
+              title="Tasks"
+              detail={describeFailure(tasks.failure)}
+              onRetry={retryAll}
             />
-          </div>
-          <TaskListView tasks={filteredTasks} onTaskClick={setSelectedTask} />
+          ) : (
+            <>
+              <div className="mb-4">
+                <TaskStatusSummary
+                  tasks={projectTasks}
+                  activeFilter={taskFilter}
+                  onFilterChange={setTaskFilter}
+                />
+              </div>
+              <TaskListView tasks={filteredTasks} onTaskClick={setSelectedTask} />
+            </>
+          )}
         </div>
       ) : null}
 
-      {activeTab === 'missions' ? <MissionTimeline missions={missions} /> : null}
+      {activeTab === 'missions' ? (
+        projectMissions === null ? (
+          <UnavailableDataNotice
+            title="Missions"
+            detail={describeFailure(missions.failure)}
+            onRetry={retryAll}
+          />
+        ) : (
+          <MissionTimeline missions={projectMissions} />
+        )
+      ) : null}
 
       {activeTab === 'prd' && prdContent ? (
         <div className="rounded-lg border border-surface-border bg-surface-card p-6">
@@ -248,18 +351,26 @@ function OverviewTab({
   tasks,
 }: {
   project: Project;
-  missions: Mission[];
-  tasks: Task[];
+  missions: Mission[] | null;
+  tasks: Task[] | null;
 }): ReactElement {
-  const recentTasks = [...tasks]
-    .sort((left, right) => new Date(right.updatedAt).getTime() - new Date(left.updatedAt).getTime())
-    .slice(0, 5);
+  const recentTasks =
+    tasks === null
+      ? null
+      : [...tasks]
+          .sort(
+            (left, right) =>
+              new Date(right.updatedAt).getTime() - new Date(left.updatedAt).getTime(),
+          )
+          .slice(0, 5);
 
   return (
     <div className="grid gap-6 lg:grid-cols-2">
       <section>
         <h2 className="mb-3 text-sm font-semibold text-text-secondary">Recent Tasks</h2>
-        {recentTasks.length === 0 ? (
+        {recentTasks === null ? (
+          <UnavailableDataNotice title="Tasks" />
+        ) : recentTasks.length === 0 ? (
           <div className="rounded-lg border border-surface-border bg-surface-card p-4 text-center">
             <p className="text-sm text-text-muted">No tasks yet</p>
           </div>
@@ -287,7 +398,9 @@ function OverviewTab({
 
       <section>
         <h2 className="mb-3 text-sm font-semibold text-text-secondary">Missions</h2>
-        {missions.length === 0 ? (
+        {missions === null ? (
+          <UnavailableDataNotice title="Missions" />
+        ) : missions.length === 0 ? (
           <div className="rounded-lg border border-surface-border bg-surface-card p-4 text-center">
             <p className="text-sm text-text-muted">No missions yet</p>
           </div>

apps/web/src/spa/pages/projects.spec.tsx

@@ -51,6 +51,7 @@ afterEach(async () => {
   document.body.replaceChildren();
   root = null;
   apiMock.mockReset();
+  sessionStorage.clear();
 });
 
 async function renderProjectsPage(): Promise<ReturnType<typeof createMemoryRouter>> {
@@ -71,6 +72,22 @@ async function renderProjectsPage(): Promise<ReturnType<typeof createMemoryRoute
   return router;
 }
 
+function clickButtonByText(text: string): void {
+  const button = [...container.querySelectorAll('button')].find((candidate) =>
+    candidate.textContent?.includes(text),
+  );
+  if (!button) {
+    throw new Error(`Button containing "${text}" not found`);
+  }
+  button.dispatchEvent(new MouseEvent('click', { bubbles: true }));
+}
+
+async function flushAct(): Promise<void> {
+  await act(async () => {
+    await Promise.resolve();
+  });
+}
+
 describe('ProjectsPage', () => {
   it('shows a visible loading state while the project request is in flight', async () => {
     const deferred = createDeferred<typeof projectFixtures>();
@@ -91,7 +108,7 @@ describe('ProjectsPage', () => {
 
     const router = await renderProjectsPage();
 
-    expect(apiMock).toHaveBeenCalledWith('/api/projects');
+    expect(apiMock.mock.calls[0]?.[0]).toBe('/api/projects');
     expect(container.textContent).toContain('Mosaic Stack');
     expect(container.textContent).toContain('Agent Runtime');
 
@@ -108,7 +125,7 @@ describe('ProjectsPage', () => {
     expect(container.textContent).toContain('Project detail target');
   });
 
-  it('renders the empty state when the API returns no projects', async () => {
+  it('renders the empty state only for a verified empty collection', async () => {
     apiMock.mockResolvedValueOnce([]);
 
     await renderProjectsPage();
@@ -117,9 +134,12 @@ describe('ProjectsPage', () => {
     expect(container.textContent).toContain(
       'Projects will appear here when created via the gateway API',
     );
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
   });
 
-  it('renders a visible alert when the projects request fails', async () => {
+  it('renders a failed fetch as an explicit unavailable state, never an empty collection', async () => {
     apiMock.mockRejectedValueOnce(new Error('Projects are unavailable'));
 
     await renderProjectsPage();
@@ -127,5 +147,51 @@ describe('ProjectsPage', () => {
     const alert = container.querySelector('[role="alert"]');
     expect(alert).toBeTruthy();
     expect(alert?.textContent).toContain('Projects are unavailable');
+    expect(alert?.textContent).toContain('not an empty result');
+
+    // Negative controls: no healthy empty state and no project cards render
+    // from a failed fetch.
+    expect(container.textContent).not.toContain('No projects yet');
+    expect(container.textContent).not.toContain('Mosaic Stack');
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'unavailable',
+    );
+  });
+
+  it('renders an auth failure as unavailable and recovers after retry', async () => {
+    apiMock
+      .mockRejectedValueOnce(Object.assign(new Error('Unauthorized'), { statusCode: 401 }))
+      .mockResolvedValueOnce(projectFixtures);
+
+    await renderProjectsPage();
+
+    const alert = container.querySelector('[role="alert"]');
+    expect(alert?.textContent).toContain('Unauthorized');
+    expect(container.textContent).not.toContain('No projects yet');
+
+    await act(async () => {
+      clickButtonByText('Retry');
+    });
+    await flushAct();
+
+    expect(container.querySelector('[role="alert"]')).toBeNull();
+    expect(container.textContent).toContain('Mosaic Stack');
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
+  });
+
+  it('renders a schema-mismatched response as unavailable, never as data', async () => {
+    apiMock.mockResolvedValueOnce({ results: projectFixtures });
+
+    await renderProjectsPage();
+
+    const alert = container.querySelector('[role="alert"]');
+    expect(alert?.textContent).toContain('not an empty result');
+    expect(container.textContent).not.toContain('Mosaic Stack');
+    expect(container.textContent).not.toContain('No projects yet');
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'unavailable',
+    );
   });
 });

apps/web/src/spa/pages/projects.tsx

@@ -1,53 +1,51 @@
-import { useEffect, useState, type ReactElement } from 'react';
+import { type ReactElement } from 'react';
 import { useNavigate } from 'react-router-dom';
 import { ProjectCard } from '@/components/projects/project-card';
+import { StaleDataNotice, UnavailableDataNotice } from '@/components/freshness/freshness-notices';
 import { api } from '@/lib/api';
 import type { Project } from '@/lib/types';
-import { getErrorMessage } from './page-errors';
+import { useFreshCollection, describeFailure } from '@/lib/freshness/use-fresh-collection';
+import { validateProjectCollection } from '@/lib/freshness/validators';
 
 export function ProjectsPage(): ReactElement {
   const navigate = useNavigate();
-  const [projects, setProjects] = useState<Project[]>([]);
-  const [loading, setLoading] = useState(true);
-  const [error, setError] = useState<string | null>(null);
-
-  useEffect(() => {
-    let cancelled = false;
-
-    void api<Project[]>('/api/projects')
-      .then((response) => {
-        if (cancelled) return;
-        setProjects(response);
-      })
-      .catch((caught: unknown) => {
-        if (cancelled) return;
-        setError(getErrorMessage(caught, 'Failed to load projects.'));
-      })
-      .finally(() => {
-        if (cancelled) return;
-        setLoading(false);
-      });
-
-    return () => {
-      cancelled = true;
-    };
-  }, []);
+  const projects = useFreshCollection<Project[]>({
+    source: 'gateway:/api/projects',
+    fetcher: (signal) => api<unknown>('/api/projects', { signal }),
+    validate: validateProjectCollection,
+    // Projects carry workspace identity (userId) that is only knowable from
+    // the payload itself, so a restored entry cannot be scope-checked before
+    // display. Conservative choice: no last-known restore for this surface;
+    // cross-workspace switching is still invalidated at verification time.
+  });
+  const retry = (): void => {
+    void projects.revalidate();
+  };
 
   return (
-    <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
+    <div
+      data-freshness={projects.freshness}
+      className="flex min-h-screen flex-col px-4 py-6 sm:px-6"
+    >
       <header className="mb-6 border-b px-1 pb-3">
         <h1 className="text-2xl font-semibold">Projects</h1>
       </header>
 
-      {error ? (
-        <div role="alert" className="mb-6 rounded-lg border border-error/40 px-4 py-3 text-sm">
-          {error}
+      {projects.freshness === 'stale' && projects.snapshot ? (
+        <div className="mb-6">
+          <StaleDataNotice label={projects.snapshot} onRetry={retry} />
         </div>
       ) : null}
 
-      {loading ? (
+      {projects.freshness === 'unknown' ? (
         <p className="py-8 text-center text-sm text-text-muted">Loading projects...</p>
-      ) : projects.length === 0 ? (
+      ) : projects.freshness === 'unavailable' ? (
+        <UnavailableDataNotice
+          title="Projects"
+          detail={describeFailure(projects.failure)}
+          onRetry={retry}
+        />
+      ) : projects.data !== null && projects.data.length === 0 ? (
         <div className="py-12 text-center">
           <h2 className="text-lg font-medium text-text-secondary">No projects yet</h2>
           <p className="mt-1 text-sm text-text-muted">
@@ -56,7 +54,7 @@ export function ProjectsPage(): ReactElement {
         </div>
       ) : (
         <div className="grid gap-4 sm:grid-cols-2 lg:grid-cols-3">
-          {projects.map((project) => (
+          {(projects.data ?? []).map((project) => (
             <ProjectCard
               key={project.id}
               project={project}

apps/web/src/spa/pages/tasks.spec.tsx

@@ -3,6 +3,9 @@ import { createRoot, type Root } from 'react-dom/client';
 import { createMemoryRouter, RouterProvider, type RouteObject } from 'react-router-dom';
 import { afterAll, afterEach, beforeAll, describe, expect, it, vi } from 'vitest';
 import { taskFixtures } from './page-fixtures';
+import { acceptSnapshot, DEFAULT_FRESHNESS_POLICY } from '@/lib/freshness/model';
+import { writeSnapshotCache } from '@/lib/freshness/snapshot-cache';
+import { validateTaskCollection } from '@/lib/freshness/validators';
 
 const { apiMock } = vi.hoisted(() => ({
   apiMock: vi.fn(),
@@ -48,6 +51,7 @@ afterEach(async () => {
   document.body.replaceChildren();
   root = null;
   apiMock.mockReset();
+  sessionStorage.clear();
 });
 
 async function renderTasksPage(): Promise<void> {
@@ -72,6 +76,13 @@ function clickButtonByText(text: string): void {
   button.dispatchEvent(new MouseEvent('click', { bubbles: true }));
 }
 
+/** Flush pending promise callbacks inside the act environment. */
+async function flushAct(): Promise<void> {
+  await act(async () => {
+    await Promise.resolve();
+  });
+}
+
 describe('TasksPage', () => {
   it('shows a visible loading state before the tasks request settles', async () => {
     const deferred = createDeferred<typeof taskFixtures>();
@@ -132,7 +143,7 @@ describe('TasksPage', () => {
     expect(container.textContent).toContain('Wire list and kanban modal interactions');
   });
 
-  it('renders a visible alert when the tasks request fails', async () => {
+  it('renders a failed fetch as an explicit unavailable state, never an empty healthy board', async () => {
     apiMock.mockRejectedValueOnce(new Error('Tasks request failed'));
 
     await renderTasksPage();
@@ -140,5 +151,80 @@ describe('TasksPage', () => {
     const alert = container.querySelector('[role="alert"]');
     expect(alert).toBeTruthy();
     expect(alert?.textContent).toContain('Tasks request failed');
+    expect(alert?.textContent).toContain('not an empty result');
+
+    // Negative controls: no board, no healthy empty-state markers, and the
+    // surface is marked unavailable rather than current.
+    expect(container.textContent).not.toContain('Not Started');
+    expect(container.textContent).not.toContain('No tasks');
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'unavailable',
+    );
+  });
+
+  it('recovers to a current board after retrying a failed fetch', async () => {
+    apiMock
+      .mockRejectedValueOnce(new Error('Tasks request failed'))
+      .mockResolvedValueOnce(taskFixtures);
+
+    await renderTasksPage();
+    expect(container.querySelector('[role="alert"]')).toBeTruthy();
+
+    await act(async () => {
+      clickButtonByText('Retry');
+    });
+    await flushAct();
+
+    expect(container.querySelector('[role="alert"]')).toBeNull();
+    expect(container.textContent).toContain('Not Started');
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
+  });
+
+  it('labels restored last-known data as stale with source, version, and age until verified', async () => {
+    // Seed a last-known snapshot fetched five minutes ago; the page must
+    // render it only under an explicit staleness label while the fetch is
+    // still in flight.
+    const restored = acceptSnapshot({
+      value: taskFixtures,
+      validate: validateTaskCollection,
+      previous: null,
+      policy: DEFAULT_FRESHNESS_POLICY,
+      source: 'gateway:/api/tasks',
+      now: Date.now() - 5 * 60_000,
+    });
+    if (restored.outcome !== 'accepted') throw new Error('fixture setup failed');
+    writeSnapshotCache('tasks', restored.snapshot);
+
+    const deferred = createDeferred<typeof taskFixtures>();
+    apiMock.mockReturnValueOnce(deferred.promise);
+
+    await renderTasksPage();
+
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'stale',
+    );
+    const banner = container.querySelector('[role="status"]');
+    expect(banner?.textContent).toContain('last-known');
+    expect(banner?.textContent).toContain('may be out of date');
+    expect(banner?.textContent).toContain('gateway:/api/tasks');
+    expect(banner?.textContent).toContain('snapshot v1');
+    expect(banner?.textContent).toContain('5m ago');
+
+    // Last-known data still renders as situational awareness under the label.
+    expect(container.textContent).toContain('Route /tasks');
+    expect(container.textContent).not.toContain('Loading tasks...');
+
+    // Verification lands: the banner clears and the surface becomes current.
+    await act(async () => {
+      deferred.resolve(taskFixtures);
+      await deferred.promise;
+    });
+
+    expect(container.querySelector('[role="status"]')).toBeNull();
+    expect(container.querySelector('[data-freshness]')?.getAttribute('data-freshness')).toBe(
+      'current',
+    );
   });
 });

apps/web/src/spa/pages/tasks.tsx

@@ -1,45 +1,32 @@
-import { useEffect, useState, type ReactElement } from 'react';
+import { useState, type ReactElement } from 'react';
 import { KanbanBoard } from '@/components/tasks/kanban-board';
 import { TaskDetailModal } from '@/components/tasks/task-detail-modal';
 import { TaskListView } from '@/components/tasks/task-list-view';
+import { StaleDataNotice, UnavailableDataNotice } from '@/components/freshness/freshness-notices';
 import { api } from '@/lib/api';
 import { cn } from '@/lib/cn';
 import type { Task } from '@/lib/types';
-import { getErrorMessage } from './page-errors';
+import { useFreshCollection, describeFailure } from '@/lib/freshness/use-fresh-collection';
+import { validateTaskCollection } from '@/lib/freshness/validators';
 
 type ViewMode = 'list' | 'kanban';
 
 export function TasksPage(): ReactElement {
-  const [tasks, setTasks] = useState<Task[]>([]);
+  const tasks = useFreshCollection<Task[]>({
+    source: 'gateway:/api/tasks',
+    fetcher: (signal) => api<unknown>('/api/tasks', { signal }),
+    validate: validateTaskCollection,
+    cacheKey: 'tasks',
+  });
   const [view, setView] = useState<ViewMode>('kanban');
-  const [loading, setLoading] = useState(true);
-  const [error, setError] = useState<string | null>(null);
   const [selectedTask, setSelectedTask] = useState<Task | null>(null);
 
-  useEffect(() => {
-    let cancelled = false;
-
-    void api<Task[]>('/api/tasks')
-      .then((response) => {
-        if (cancelled) return;
-        setTasks(response);
-      })
-      .catch((caught: unknown) => {
-        if (cancelled) return;
-        setError(getErrorMessage(caught, 'Failed to load tasks.'));
-      })
-      .finally(() => {
-        if (cancelled) return;
-        setLoading(false);
-      });
-
-    return () => {
-      cancelled = true;
-    };
-  }, []);
+  const retry = (): void => {
+    void tasks.revalidate();
+  };
 
   return (
-    <div className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
+    <div data-freshness={tasks.freshness} className="flex min-h-screen flex-col px-4 py-6 sm:px-6">
       <header className="mb-6 flex items-center justify-between gap-4 border-b px-1 pb-3">
         <h1 className="text-2xl font-semibold">Tasks</h1>
         <div className="flex rounded-lg border border-surface-border">
@@ -70,18 +57,24 @@ export function TasksPage(): ReactElement {
         </div>
       </header>
 
-      {error ? (
-        <div role="alert" className="mb-6 rounded-lg border border-error/40 px-4 py-3 text-sm">
-          {error}
+      {tasks.freshness === 'stale' && tasks.snapshot ? (
+        <div className="mb-6">
+          <StaleDataNotice label={tasks.snapshot} onRetry={retry} />
         </div>
       ) : null}
 
-      {loading ? (
+      {tasks.freshness === 'unknown' ? (
         <p className="py-8 text-center text-sm text-text-muted">Loading tasks...</p>
+      ) : tasks.freshness === 'unavailable' ? (
+        <UnavailableDataNotice
+          title="Tasks"
+          detail={describeFailure(tasks.failure)}
+          onRetry={retry}
+        />
       ) : view === 'kanban' ? (
-        <KanbanBoard tasks={tasks} onTaskClick={setSelectedTask} />
+        <KanbanBoard tasks={tasks.data ?? []} onTaskClick={setSelectedTask} />
       ) : (
-        <TaskListView tasks={tasks} onTaskClick={setSelectedTask} />
+        <TaskListView tasks={tasks.data ?? []} onTaskClick={setSelectedTask} />
       )}
 
       {selectedTask ? (

docs/PRD.md

@@ -1614,3 +1614,38 @@ will route around it.
 10. ASSUMPTION: **Conversations and messages get their own PG tables** (not stored in brain's entity model). They follow a chat-specific schema with proper foreign keys to users and projects. Rationale: Chat has different access patterns (streaming, pagination, search) than brain entities.
 
 11. RESOLVED: **Pi handles all target LLM providers natively.** Anthropic, OpenAI/Codex, Z.ai, Ollama, LM Studio, and llama.cpp are all supported via Pi's built-in providers or `models.json` configuration with `openai-completions` API type. No custom provider adapters needed in @mosaicstack/agent — only configuration management.
+
+---
+
+## Release Integrity Workstream (RI, #1275)
+
+### Problem and objective
+
+At `next` 476db12b (review of 2026-08-17), publication from `next` is not bound to the full verification pipeline for the same commit: the publish pipeline's publish steps depend on `build` only, while ordinary push CI excludes `next`. Public Forge/MACP paths contain false-success placeholders: a stub executor that reports `completed` with exit zero, planning/remediation gates that execute literal `true`, a review gate that echoes an approving verdict, and a gate runner that treats empty commands and unimplemented CI-provider gates as passing. Shipping UI surfaces can render a failed fetch as an empty, healthy collection.
+
+Objective: for alpha 0.0.50, the release cannot publish, report, or display work state that the repository has not actually verified. Decisions SDLC-D-033 through SDLC-D-038 (Jason, 2026-08-17) scope this floor; full decision text and required-behavior lists live in jarvis-brain `docs/plans/2026-08-16_mosaic-stack-sdlc-protocol.md` and `data/decisions/mosaic-stack-sdlc-protocol.json`. This section restates only the normative requirements.
+
+### Normative requirements
+
+1. **RI-N1 Exact-commit publication verification (SDLC-D-034).** One canonical terminal verification command performs self-contained re-verification in the publish pipeline against the job's checked-out commit before any external publication effect. The command contains or invokes the complete mandatory verification set (semantic parity with the PR merge gate, including sanitization, upgrade-guard, typecheck, lint, format check, tests, and build); CI and publication do not maintain separate semantic checklists. Every publish step depends on the verification step in the executable pipeline DAG. Provider commit identity and `git rev-parse HEAD` must identify the same commit. Missing, skipped, cancelled, stale, or inconclusive checks fail closed. Documentation-only runs may skip publication but cannot bypass verification when a publication effect will occur. A negative control must prove that a broken check blocks every publish step.
+
+2. **RI-N2 Fail-closed Forge/MACP with explicit simulation (SDLC-D-035).** Simulation requires explicit caller intent (e.g. `--simulate`) and produces a distinct typed `simulated` state that can never satisfy dependencies, acceptance criteria, gates, merge, or release. Normal execution exits nonzero with a typed capability failure when a required executor, reviewer, command, or CI provider is absent — no stub completion, no literal-`true` gates, no synthetic approvals, no empty-command passes. A manual gate with no automation enters a waiting state; it does not pass. Positive tests prove explicit simulation still works; negative controls prove simulation and every missing-provider case cannot advance lifecycle state.
+
+3. **RI-N3 One transitional PRD authority (SDLC-D-036).** `@mosaicstack/prdy` structured storage under `docs/prdy/`, driven by `mosaic mission --plan`, is the authoritative PRD representation for the alpha. `mosaic prdy` either routes through the same application service or operates only as an explicit, named Markdown import/export adapter; `docs/PRD.md` is not a peer authority. `mission --plan` must persist the mission↔PRD linkage (mission id/version, PRD id/version, selected requirements). Markdown output is a generated view carrying source identity; editing it cannot mutate authority silently. Import is explicit, validated, and conflict-aware (proposed successor, never overwrite). Structural validity is separate from approval.
+
+4. **RI-N4 One quality-rails evaluator (SDLC-D-037).** The TypeScript quality-rails package is the sole authoritative evaluator. A complete probe inventory maps every current TypeScript and shell check to one canonical check with disposition (preserve/strengthen/retire, each named). Effective shell enforcement probes are absorbed before their independent paths retire; expected-file presence alone is not parity. The evaluator returns typed results (`passed`/`failed`/`blocked`/`error`/`not-applicable`) with check version, subject, and reason; missing implementation, missing input, unknown check, process error, timeout, or malformed output can never become `passed` or an unqualified skip. Check definitions and policy are versioned and digested. Shell commands become thin adapters with no separate verdict logic. The canonical terminal verification command (RI-N1) invokes this evaluator rather than duplicating its logic. Contract, parity, and negative-control tests are required, plus independent review of probe equivalence.
+
+5. **RI-N5 Consequence-aware stale UI (SDLC-D-038).** Mission Control distinguishes typed freshness states (`current`, `stale`, `partial`, `unknown`, `unavailable`) rather than inferring from empty arrays or null. A failed fetch never renders as an empty healthy collection. Last-known data may display for situational awareness only with source identity, version, and age visibly labeled; any derived completion/assurance/release verdict whose inputs are stale becomes `unknown`; all state-changing actions are disabled until fresh state loads and is revalidated. With no verified snapshot, surfaces show an explicit unavailable state. Cache corruption, cross-workspace data, schema mismatch, and version regression invalidate the snapshot. Tests cover the failure matrix (network, auth, malformed, partial, corruption, stale age, schema mismatch, recovery, stale-action rejection) with negative controls proving no case yields a current green verdict or enabled mutation.
+
+### Acceptance criteria
+
+- AC-RI-1: A push to `next` that fails any mandatory verification step publishes nothing (no npm package, no image), demonstrated by a checked-in negative control and by pipeline evidence on a real `next` publish run where the verification step is green and every publish step depends on it.
+- AC-RI-2: With no executor/reviewer/CI provider wired, Forge and MACP normal runs exit nonzero with typed capability failures; with `--simulate`, runs complete but every result is typed `simulated` and cannot satisfy any gate, dependency, or completion state — proven by unit tests including negative controls.
+- AC-RI-3: A PRD created or revised through either `mosaic mission --plan` or `mosaic prdy` resolves to one authority under `docs/prdy/` with stable identities and versions; the mission↔PRD linkage survives restart; a Markdown export is labeled as generated and cannot silently become a second writer; divergent legacy content blocks baseline claims until explicitly resolved — proven by contract tests.
+- AC-RI-4: `quality-rails check` through any entry point (TS CLI, framework shell adapter) returns the same typed verdict for the same subject; the probe inventory names every legacy check's disposition; a deliberately broken probe fails closed — proven by contract/parity/negative-control tests and independent review of probe equivalence.
+- AC-RI-5: No shipping surface renders a failed fetch as an empty healthy state; stale/partial/unavailable states are typed, labeled, and mutation-disabled — proven by the failure-matrix tests.
+- AC-RI-6: All cards merged to `next` via squash PR with terminal-green CI; release evidence for 0.0.50 records commit, verification run, and published artifacts.
+
+### Out of scope
+
+The canonical dispatcher/control-plane vertical slice (work graph, execution attempts, fenced leases, typed check-in, independent verifier dispatch) is decided post-alpha (SDLC-D-033, option B). Multi-pipeline verification certificates (SDLC-D-034 option B) are post-alpha. Full AF-1..AF-4 objective matrices and Mission Control portfolio surfaces are post-alpha.

docs/SITEMAP.md

@@ -19,15 +19,6 @@
 - [MVP mission manifest](MISSION-MANIFEST.md) — control-plane mission rollup; activity and status remain under its authorized owner.
 - [Documentation catalog and truth audit](reports/documentation/2026-08-10-docs-catalog-audit.md) — complete baseline inventory, evidence labels, broken-link clusters, and migration recommendations.
 
-## Pi persistent goals
-
-- [Persistent goal user guide](_old_structure/guides/user-guide.md#pi-persistent-goals) — `/goal` commands, verification behavior, limits, compaction/resume semantics, and limitations.
-- [Goal extension developer guide](guides/dev-guide.md#pi-persistent-goal-extension) — framework ownership, launcher ordering, lifecycle design, tests, and local Mosaic-path smoke workflow.
-- [Goal loop operations](_old_structure/guides/admin-guide.md#pi-goal-loop-operations) — deployment ownership, bounded settings, pause/resume procedures, and supervisor boundary.
-- [Pi runtime reference](../packages/mosaic/framework/runtime/pi/RUNTIME.md#extensions) — deployed paths, command summary, and bounded environment settings.
-
-## Fleet configuration management
-
 ## Protected current authority and executable books
 
 These paths remain canonical because current source/tests consume them or because the KBN authority process protects them. Relocation requires an explicitly coordinated authority and consumer migration, not documentation-only cleanup.

docs/TASKS.md

@@ -1,5 +1,19 @@
 # Tasks — MVP (Top-Level Rollup)
 
+> ---
+>
+> **STATUS: SUPERSEDED — 2026-08-20.** kind `tracking` · superseded by `docs/fleet/NORTH_STAR.yaml`
+>
+> This file is the pre-backlog tracking mechanism. `NS-2` in the north star declares the
+> replacement: every backlog item is a Mosaic Backlog card projected from the YAML. That
+> model replaced this one and nobody retired the old file, so it kept reading as
+> authoritative while going stale.
+>
+> **Do not trust a status in this file.** Verified 2026-08-20: it was already behind the
+> code when it froze five weeks ago.
+>
+> Kept as a record of what was believed. Do not update it; update the YAML.
+
 > Single-writer: orchestrator only. Workers read but never modify.
 >
 > **Mission:** mvp-20260312
@@ -108,7 +122,7 @@ Active workstream is **W1 — Federation v1**. Workers should:
 
 ## north-star doctrine consolidation — doc PR — feat/north-star-doctrine
 
-- Status: applied Mos's consolidated merge-map to docs/fleet/north-star.md (budget governance + control plane/central register + 200k cap + delegation + unified-identity Fleet + role-based naming + tmux security + drift re-captures). Doctrine only; #622/#623/#625/#628 out-of-scope. Conflict checklist green. Detail: scratchpads/north-star-doctrine.md.
+- Status: applied Mos's consolidated merge-map to docs/fleet/FLEET-DOCTRINE.md (budget governance + control plane/central register + 200k cap + delegation + unified-identity Fleet + role-based naming + tmux security + drift re-captures). Doctrine only; #622/#623/#625/#628 out-of-scope. Conflict checklist green. Detail: scratchpads/north-star-doctrine.md.
 
 ## #631 — re-seed preserves user fleet data (CRITICAL) — fix/631-reseed-preserves-fleet-data
 

docs/federation/TASKS.md

@@ -1,5 +1,21 @@
 # Tasks — Federation v1
 
+> ---
+>
+> **STATUS: SUPERSEDED — 2026-08-20.** kind `tracking` · superseded by `docs/fleet/NORTH_STAR.yaml`
+>
+> This file is the pre-backlog tracking mechanism. `NS-2` in the north star declares the
+> replacement: every backlog item is a Mosaic Backlog card projected from the YAML. That
+> model replaced this one and nobody retired the old file, so it kept reading as
+> authoritative while going stale.
+>
+> **Do not trust a status in this file.** Verified 2026-08-20: it was already behind the
+> code when it froze five weeks ago. `FED-M3-06` is marked not-started and `get.controller.ts`
+> has existed for eight weeks; `FED-M3-10/11` claim no tests exist while fifteen spec files
+> do. `FED-M2-DEPLOY-IMG-FIX` names `apps/gateway/Dockerfile`, which is not in the repo.
+>
+> Kept as a record of what was believed. Do not update it; update the YAML.
+
 > Single-writer: orchestrator only. Workers read but never modify.
 >
 > **Mission:** federation-v1-20260419

docs/fleet/FLEET-DOCTRINE.md

@@ -1,9 +1,26 @@
-# Mosaic Fleet — North Star
+---
+kind: spec
+parent: docs/fleet/NORTH_STAR.yaml
+status: active
+---
 
+# Mosaic Fleet — Doctrine
+
+> **This is the WHY. `NORTH_STAR.yaml` is the WHAT and WHEN.**
+> Renamed from `north-star.md` on 2026-08-20. It sat one character away from the
+> generated `NORTH_STAR.md` in the same directory, and the two are read by different
+> populations — the PRDs and TASKS files cite this one, while the agent role contracts
+> and the generator spec cite the YAML pair. Same-name-different-thing was the confusion;
+> the content was never in conflict.
+>
+> **Nothing here overrides `NORTH_STAR.yaml`.** Where this document states a plan item,
+> the YAML is authoritative. Where it states a decision, a rationale, or a role
+> definition, this document is the record and the YAML carries none of it.
+>
 > **Workstream:** W-FLEET (Fleet) under mission `mvp-20260312`
-> **Umbrella:** [docs/MISSION-MANIFEST.md](../MISSION-MANIFEST.md) · [docs/PRD.md](../PRD.md) (Mosaic Stack v0.1.0)
-> **Status:** doctrine — authored 2026-06-20. Owner of this file: Fleet workstream lead.
-> This document does **not** modify the MVP rollup; a rollup row is proposed, not written here.
+> **Umbrella:** [docs/MISSION-MANIFEST.md](../MISSION-MANIFEST.md)
+> **Authored:** 2026-06-20. Owner: Fleet workstream lead.
+> This document does **not** modify the MVP rollup.
 
 ## Vision
 
@@ -264,15 +281,17 @@ Dedicated Postgres **instance** vs. dedicated **schema** in the existing instanc
 Recommendation: dedicated schema, existing instance (a migration file, not new infra);
 re-evaluate if isolation or write-volume demands it.
 
-## Phased roadmap
+## Phased roadmap — SUPERSEDED
 
-| Phase                  | Outcome                                                                                                                                                                                                | Status  |
-| ---------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ------- |
-| 0–1                    | tmux PoC, hardening, published CLI v0.0.34 (#565–#568)                                                                                                                                                 | ✅ done |
-| **2 — Observability**  | fleet ps (host+tenant aware join), heartbeat protocol + dogfood stub answers it, agent watch (read-only), agent send --verify receipts                                                                 | ▶ now   |
-| 3 — Real runtimes      | claude/codex/pi/opencode answer heartbeat; **hybrid lifecycle** (core always-on: **orchestrator + enhancer**; ephemeral workers per lane)                                                              | planned |
-| 4 — Unified definition | one agent schema in gateway; mosaic agent --new → materialized per-tenant session; uid-tenant provisioning; **`fleet` schema migration + `forge-exec` TaskExecutor adapter (forge → `agent-send.sh`)** | planned |
-| 5 — Control plane      | federation-backed cross-host × cross-tenant fleet view; **webUI** (surface chosen then) for MVP-X1 parity; **central register live (spend ledger, docs-as-projections, multi-host Kanban)**            | planned |
+Superseded 2026-08-20 by [`NORTH_STAR.yaml`](./NORTH_STAR.yaml), whose `goals` carry both
+a `phase` (build order) and a `tier` (which promise the goal delivers). The five-phase
+table that stood here could not express those as separate axes, and its "Phase 2 —
+Observability ▶ now" row stayed unfalsified for two months because a phase has no exit
+test. Tiers do: see `AC-NS-0` through `AC-NS-7`.
+
+The phase-2 content itself is not lost — it is specified in
+[`PRD.md`](./PRD.md) (Fleet Phase 2: Operator Observability) and is now tracked as
+goals `I1`–`I5` at tier 0.
 
 ## Decisions of record (2026-06-20, with Jason)
 

docs/fleet/NORTH_STAR.md

@@ -3,7 +3,7 @@
 > **Generated file — do not edit by hand.**
 > Projected deterministically from [`NORTH_STAR.yaml`](./NORTH_STAR.yaml) by the pure
 > generator in `packages/mosaic/src/commands/fleet.ts` (`renderNorthStarMarkdown`).
-> Edit the YAML, then regenerate. Self-contained Mosaic — no Hermes dependency.
+> Edit the YAML, then regenerate. Self-contained Mosaic.
 
 ## Mission
 
@@ -11,7 +11,7 @@ A self-driving Mosaic system that 24/7 unattended converts a machine-readable go
 
 ## Substrate
 
-The Mosaic Backlog is the backlog of record + dispatch engine, built on Mosaic's native Postgres storage service (@mosaicstack/db drizzle; PGlite-embedded by default, full Postgres by config). NOT Hermes.
+The Mosaic Backlog is the backlog of record + dispatch engine, built on Mosaic's native Postgres storage service (@mosaicstack/db drizzle; PGlite-embedded by default, full Postgres by config).
 
 ## Standing objectives
 
@@ -24,16 +24,18 @@ The Mosaic Backlog is the backlog of record + dispatch engine, built on Mosaic's
 - **NS-7** — Meta-loop (session-review + enhancer) continuously proposes small fleet-improvement PRs.
 - **NS-8** — Single operator-flippable PAUSE kill-switch (fleet/run/PAUSED) honored before every dispatch and every merge.
 - **NS-9** — Mosaic is a general-purpose multi-agent system: the user declares the SYSTEM TYPE to run (e.g. software delivery, personal assistant, research, business/operations) and the orchestrator provisions the matching persona roster and org structure from a cross-domain baseline persona library; the delivery/coding fleet is one profile among many.
+- **NS-10** — An adoption is not complete until the mechanism it replaces is removed. Two live conventions for one concern is the defect, not a transition state. Measured 2026-08-20: brain-home adopted by 9 modules and not 10; MOSAIC_HOME honored in 4 places, each re-deriving it; backlog cards declared while TASKS.md files stayed authoritative. Every one was decided correctly and left half-applied.
 
 ## Success criteria
 
-- **AC-NS-1** — The supervisor keeps a two-agent floor (1 orchestrator + >=1 enhancer) healthy across reboot.
-- **AC-NS-2** — A goal added to this YAML is decomposed to cards and either merged or escalated, with no human in the loop.
-- **AC-NS-3** — No PR merges with failure/error/no-status/timeout CI, and none bypass pr-merge.sh.
-- **AC-NS-4** — TTL is enforced on claims; token caps remain advisory until a real meter exists.
-- **AC-NS-5** — Flipping fleet/run/PAUSED halts dispatch and merges within one tick.
-- **AC-NS-6** — A user can declare a system type and the fleet provisions the matching persona roster + topology from the baseline library, with no code change.
-- **AC-NS-7** — A user-customized persona (edited or added via the orchestrator) survives mosaic update: baseline reseed never clobbers user overrides.
+- **AC-NS-0** (tier 0) — The operator launches an agent on any configured harness with one command, observes its state and sends it work without attaching to a terminal multiplexer.
+- **AC-NS-1** (tier 1) — The supervisor keeps a two-agent floor (1 orchestrator + >=1 enhancer) healthy across reboot.
+- **AC-NS-2** (tier 1) — A goal added to this YAML is decomposed to cards and either merged or escalated, with no human in the loop.
+- **AC-NS-3** (tier 1) — No PR merges with failure/error/no-status/timeout CI, and none bypass pr-merge.sh.
+- **AC-NS-4** (tier 1) — TTL is enforced on claims; token caps remain advisory until a real meter exists.
+- **AC-NS-5** (tier 1) — Flipping fleet/run/PAUSED halts dispatch and merges within one tick.
+- **AC-NS-6** (tier 2) — A user can declare a system type and the fleet provisions the matching persona roster + topology from the baseline library, with no code change.
+- **AC-NS-7** (tier 2) — A user-customized persona (edited or added via the orchestrator) survives mosaic update: baseline reseed never clobbers user overrides.
 
 ## Workstreams
 
@@ -45,26 +47,44 @@ The Mosaic Backlog is the backlog of record + dispatch engine, built on Mosaic's
 | D   | Merge-gate — single approver, pr-merge.sh after CI wait                                                     |
 | E   | Meta-loop — session-review + enhancer improvement PRs                                                       |
 | F   | Safety-rails — TTL claims, advisory spend, PAUSE kill-switch                                                |
+| G   | Kill-switch — operator PAUSE honored before dispatch and merge                                              |
 | H   | Personas & system profiles — cross-domain library, system-type provisioning, update-surviving customization |
+| I   | Operator surface — launcher, fleet visibility, reliable steering (tier 0)                                   |
+| J   | Web control plane — browser surface over the gateway (tier 1)                                               |
+| K   | Clients — desktop and mobile over the same backend (tier 2)                                                 |
+| L   | Auth profiles — per-provider accounts, per-session selection (tier 2)                                       |
 
 ## Goals (backlog projection)
 
-| id  | title                                                                                                                                             | phase | priority    | depends_on |
-| --- | ------------------------------------------------------------------------------------------------------------------------------------------------- | ----- | ----------- | ---------- |
-| A1  | Machine-readable NORTH_STAR.yaml + Markdown projection                                                                                            | 1     | must-have   | —          |
-| A2  | Mosaic Backlog schema + storage-service card store (drizzle/PGlite)                                                                               | 1     | must-have   | A1         |
-| A3a | Card lifecycle — create/claim/release with stable ids + depends_on DAG                                                                            | 1     | must-have   | A2         |
-| A3b | TTL-bounded claim enforcement (wall-clock) on cards                                                                                               | 1     | must-have   | A3a        |
-| A4  | Advisory spend projection per card (degrades to TTL, no real meter)                                                                               | 1     | should-have | A3a        |
-| B1  | Supervisor tick — readiness scan, two-agent-floor health check                                                                                    | 2     | must-have   | A3a        |
-| B2  | Native dispatch/claim — assign ready dependency-satisfied work                                                                                    | 2     | must-have   | A3b, B1    |
-| B3a | Planner decompose — goal added to YAML → cards                                                                                                    | 2     | must-have   | A2, B1     |
-| B3b | Replan request on empty backlog; escalate on no-decompose                                                                                         | 2     | should-have | B3a        |
-| G1  | PAUSE kill-switch + merge-gate honored before dispatch and merge                                                                                  | 2     | must-have   | B2         |
-| H1  | Cross-domain baseline persona library (exec, marketing, ops, research, assistant + engineering roles)                                             | 1     | must-have   | A1         |
-| H2  | System-type profiles — declarative mapping of system type to persona roster + topology                                                            | 2     | must-have   | H1         |
-| H3  | System-type provisioning — user declares type; orchestrator instantiates the matching roster + structure                                          | 2     | must-have   | H2         |
-| H4  | Update-surviving persona customization — ad-hoc edits/additions persisted in a PRESERVE-protected override layer (baseline merged with overrides) | 2     | must-have   | H1         |
+| id  | title                                                                                                                                                                                                                                                                                                                                                                              | tier | phase | priority    | depends_on     |
+| --- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---- | ----- | ----------- | -------------- |
+| A1  | Machine-readable NORTH_STAR.yaml + Markdown projection                                                                                                                                                                                                                                                                                                                             | 1    | 1     | must-have   | —              |
+| A2  | Mosaic Backlog schema + storage-service card store (drizzle/PGlite)                                                                                                                                                                                                                                                                                                                | 1    | 1     | must-have   | A1             |
+| A3a | Card lifecycle — create/claim/release with stable ids + depends_on DAG                                                                                                                                                                                                                                                                                                             | 1    | 1     | must-have   | A2             |
+| A3b | TTL-bounded claim enforcement (wall-clock) on cards                                                                                                                                                                                                                                                                                                                                | 1    | 1     | must-have   | A3a            |
+| A4  | Advisory spend projection per card (degrades to TTL, no real meter)                                                                                                                                                                                                                                                                                                                | 1    | 1     | should-have | A3a            |
+| B1  | Supervisor tick — readiness scan, two-agent-floor health check                                                                                                                                                                                                                                                                                                                     | 1    | 2     | must-have   | A3a            |
+| B2  | Native dispatch/claim — assign ready dependency-satisfied work                                                                                                                                                                                                                                                                                                                     | 1    | 2     | must-have   | A3b, B1        |
+| B3a | Planner decompose — goal added to YAML → cards                                                                                                                                                                                                                                                                                                                                     | 1    | 2     | must-have   | A2, B1         |
+| B3b | Replan request on empty backlog; escalate on no-decompose                                                                                                                                                                                                                                                                                                                          | 1    | 2     | should-have | B3a            |
+| G1  | PAUSE kill-switch + merge-gate honored before dispatch and merge                                                                                                                                                                                                                                                                                                                   | 1    | 2     | must-have   | B2             |
+| H1  | Cross-domain baseline persona library (exec, marketing, ops, research, assistant + engineering roles)                                                                                                                                                                                                                                                                              | 2    | 1     | must-have   | A1             |
+| H2  | System-type profiles — declarative mapping of system type to persona roster + topology                                                                                                                                                                                                                                                                                             | 2    | 2     | must-have   | H1             |
+| H3  | System-type provisioning — user declares type; orchestrator instantiates the matching roster + structure                                                                                                                                                                                                                                                                           | 2    | 2     | must-have   | H2             |
+| H4  | Update-surviving persona customization — ad-hoc edits/additions persisted in a PRESERVE-protected override layer (baseline merged with overrides)                                                                                                                                                                                                                                  | 2    | 2     | must-have   | H1             |
+| A5  | NORTH_STAR schema validator — every goal's workstream declared, every workstream has a goal, every depends_on id exists, every tier has a success criterion; runs in CI beside the Markdown regeneration check                                                                                                                                                                     | 0    | 1     | must-have   | A1             |
+| I1  | One home resolver — a single function resolving MOSAIC_HOME with a sane default, adopted by every module. Today brain-home.ts is imported by 9 modules while 10 still use DEFAULT_MOSAIC_HOME, and MOSAIC_HOME is re-derived ad hoc in 4 places. NS-10 applies - finish the adoption and delete the second path                                                                    | 0    | 1     | must-have   | —              |
+| I2  | mosaic fleet ps sees the fleet that is actually running. Three measured blockers: the roster declares socket `mosaic-fleet` which does not exist, the 18 live sessions are on the default socket, and nothing writes a roster because seats are launched outside the CLI. Make the socket configurable and the roster written at launch, or make ps read tmux + seat dirs directly | 0    | 1     | must-have   | I1             |
+| I3  | Migrate fleet steering onto mosaic agent send --verify (exists, FLEET-OBS-005, spec FR-5) and retire tools/tmux/agent-send.sh, which forges the sender (D33) and returns an uninformative rc (D16, D34). FR-5 predates those defects by a month                                                                                                                                    | 0    | 1     | must-have   | I1             |
+| I4  | mosaic fleet absorbs what launch-seat.sh does and launch-seat.sh is deprecated: compose the prompt file set, force the skill set, wire the style hook, fail closed on any unreadable input, export per-seat git identity, and register the seat in the roster. launch-seat.sh was a manual method; it is the reference implementation, not the destination                         | 0    | 1     | must-have   | I1, I5         |
+| I5  | Harness probe matrix — verify a working prompt-injection path for claude, codex, opencode and pi, and refuse any runtime whose path is unverified. AC-NS-0 clause 1 ("any configured harness") rests on this. The probe work in docs/plans/2026-08-19_launch-seat-multi-runtime.md (brain, untracked) transfers; its launch-seat.sh target does not                                | 0    | 1     | must-have   | —              |
+| I6  | Finish the heartbeat responder (FLEET-OBS-002, the only Phase-2 task still in-progress; spec FR-2). Health must mean "answered a heartbeat", not "pane alive" — pane state measured two seats wrong on 2026-08-20                                                                                                                                                                  | 0    | 1     | must-have   | —              |
+| I7  | Independent review and live-fleet dogfood of the Phase-2 verbs (FLEET-OBS-008), then land them (FLEET-OBS-009). Implementation is done and verification is not; "done" in a task file frozen five weeks is not evidence                                                                                                                                                            | 0    | 1     | must-have   | I2, I3, I4, I6 |
+| I8  | Neutralize misleading documentation — supersede headers on docs that state a stale status, a false blocker or a retired mechanism. Cheap, and it is tier 0 because a stale doc does not merely fail to help an agent, it actively misroutes one. Rebuilding the documentation is a separate and later job                                                                          | 0    | 1     | must-have   | —              |
+| I9  | Study t3code's agent-attach and multi-provider auth methods and record what transfers. Reference only — Mosaic implements its own within the stack, never adopts the code and never takes the dependency. Informs HOW I/J/K/L are built, not whether                                                                                                                               | 0    | 1     | should-have | —              |
+| J1  | Web control plane over the gateway — fleet visibility and steering in a browser, same data source as I2                                                                                                                                                                                                                                                                            | 1    | 3     | must-have   | I2             |
+| K1  | Desktop and mobile clients against the gateway, authenticated                                                                                                                                                                                                                                                                                                                      | 2    | 4     | must-have   | J1             |
+| L1  | Per-provider auth profiles with per-session selection                                                                                                                                                                                                                                                                                                                              | 2    | 4     | must-have   | I4             |
 
 ## Assumptions (vetoable)
 

docs/fleet/NORTH_STAR.yaml

@@ -6,7 +6,7 @@
 # packages/mosaic/src/commands/fleet.ts (renderNorthStarMarkdown). Edit the YAML,
 # never the .md.
 #
-# Self-contained Mosaic. NO Hermes runtime dependency. The backlog of record is
+# Self-contained Mosaic. The backlog of record is
 # the Mosaic Backlog on Mosaic's OWN native Postgres storage service.
 
 version: 1
@@ -24,7 +24,7 @@ substrate:
   note: >-
     The Mosaic Backlog is the backlog of record + dispatch engine, built on
     Mosaic's native Postgres storage service (@mosaicstack/db drizzle;
-    PGlite-embedded by default, full Postgres by config). NOT Hermes.
+    PGlite-embedded by default, full Postgres by config).
 
 standing_objectives:
   - id: NS-1
@@ -69,32 +69,53 @@ standing_objectives:
       business/operations) and the orchestrator provisions the matching persona
       roster and org structure from a cross-domain baseline persona library; the
       delivery/coding fleet is one profile among many.
+  - id: NS-10
+    text: >-
+      An adoption is not complete until the mechanism it replaces is removed.
+      Two live conventions for one concern is the defect, not a transition
+      state. Measured 2026-08-20: brain-home adopted by 9 modules and not 10;
+      MOSAIC_HOME honored in 4 places, each re-deriving it; backlog cards
+      declared while TASKS.md files stayed authoritative. Every one was decided
+      correctly and left half-applied.
 
 success_criteria:
+  - id: AC-NS-0
+    tier: 0
+    text: >-
+      The operator launches an agent on any configured harness with one
+      command, observes its state and sends it work without attaching to a
+      terminal multiplexer.
   - id: AC-NS-1
+    tier: 1
     text: >-
       The supervisor keeps a two-agent floor (1 orchestrator + >=1 enhancer)
       healthy across reboot.
   - id: AC-NS-2
+    tier: 1
     text: >-
       A goal added to this YAML is decomposed to cards and either merged or
       escalated, with no human in the loop.
   - id: AC-NS-3
+    tier: 1
     text: >-
       No PR merges with failure/error/no-status/timeout CI, and none bypass
       pr-merge.sh.
   - id: AC-NS-4
+    tier: 1
     text: >-
       TTL is enforced on claims; token caps remain advisory until a real meter
       exists.
   - id: AC-NS-5
+    tier: 1
     text: >-
       Flipping fleet/run/PAUSED halts dispatch and merges within one tick.
   - id: AC-NS-6
+    tier: 2
     text: >-
       A user can declare a system type and the fleet provisions the matching
       persona roster + topology from the baseline library, with no code change.
   - id: AC-NS-7
+    tier: 2
     text: >-
       A user-customized persona (edited or added via the orchestrator) survives
       mosaic update: baseline reseed never clobbers user overrides.
@@ -112,80 +133,186 @@ workstreams:
     title: Meta-loop — session-review + enhancer improvement PRs
   - id: F
     title: Safety-rails — TTL claims, advisory spend, PAUSE kill-switch
+  - id: G
+    title: Kill-switch — operator PAUSE honored before dispatch and merge
   - id: H
     title: Personas & system profiles — cross-domain library, system-type provisioning, update-surviving customization
+  - id: I
+    title: Operator surface — launcher, fleet visibility, reliable steering (tier 0)
+  - id: J
+    title: Web control plane — browser surface over the gateway (tier 1)
+  - id: K
+    title: Clients — desktop and mobile over the same backend (tier 2)
+  - id: L
+    title: Auth profiles — per-provider accounts, per-session selection (tier 2)
+
+# NOTE: workstreams C, D, E and F are declared but currently project no goals.
+# That is planning debt, not an editing error: their goals have not been written
+# yet. The A5 validator below reports it rather than letting it stay invisible.
 
 goals:
   - id: A1
     title: Machine-readable NORTH_STAR.yaml + Markdown projection
     phase: 1
+    tier: 1
     priority: must-have
     depends_on: []
   - id: A2
     title: Mosaic Backlog schema + storage-service card store (drizzle/PGlite)
     phase: 1
+    tier: 1
     priority: must-have
     depends_on: [A1]
   - id: A3a
     title: Card lifecycle — create/claim/release with stable ids + depends_on DAG
     phase: 1
+    tier: 1
     priority: must-have
     depends_on: [A2]
   - id: A3b
     title: TTL-bounded claim enforcement (wall-clock) on cards
     phase: 1
+    tier: 1
     priority: must-have
     depends_on: [A3a]
   - id: A4
     title: Advisory spend projection per card (degrades to TTL, no real meter)
     phase: 1
+    tier: 1
     priority: should-have
     depends_on: [A3a]
   - id: B1
     title: Supervisor tick — readiness scan, two-agent-floor health check
     phase: 2
+    tier: 1
     priority: must-have
     depends_on: [A3a]
   - id: B2
     title: Native dispatch/claim — assign ready dependency-satisfied work
     phase: 2
+    tier: 1
     priority: must-have
     depends_on: [A3b, B1]
   - id: B3a
     title: Planner decompose — goal added to YAML → cards
     phase: 2
+    tier: 1
     priority: must-have
     depends_on: [A2, B1]
   - id: B3b
     title: Replan request on empty backlog; escalate on no-decompose
     phase: 2
+    tier: 1
     priority: should-have
     depends_on: [B3a]
   - id: G1
     title: PAUSE kill-switch + merge-gate honored before dispatch and merge
     phase: 2
+    tier: 1
     priority: must-have
     depends_on: [B2]
   - id: H1
     title: Cross-domain baseline persona library (exec, marketing, ops, research, assistant + engineering roles)
     phase: 1
+    tier: 2
     priority: must-have
     depends_on: [A1]
   - id: H2
     title: System-type profiles — declarative mapping of system type to persona roster + topology
     phase: 2
+    tier: 2
     priority: must-have
     depends_on: [H1]
   - id: H3
     title: System-type provisioning — user declares type; orchestrator instantiates the matching roster + structure
     phase: 2
+    tier: 2
     priority: must-have
     depends_on: [H2]
   - id: H4
     title: Update-surviving persona customization — ad-hoc edits/additions persisted in a PRESERVE-protected override layer (baseline merged with overrides)
     phase: 2
+    tier: 2
     priority: must-have
     depends_on: [H1]
+  - id: A5
+    title: NORTH_STAR schema validator — every goal's workstream declared, every workstream has a goal, every depends_on id exists, every tier has a success criterion; runs in CI beside the Markdown regeneration check
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: [A1]
+  - id: I1
+    title: One home resolver — a single function resolving MOSAIC_HOME with a sane default, adopted by every module. Today brain-home.ts is imported by 9 modules while 10 still use DEFAULT_MOSAIC_HOME, and MOSAIC_HOME is re-derived ad hoc in 4 places. NS-10 applies - finish the adoption and delete the second path
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: []
+  - id: I2
+    title: 'mosaic fleet ps sees the fleet that is actually running. Three measured blockers: the roster declares socket `mosaic-fleet` which does not exist, the 18 live sessions are on the default socket, and nothing writes a roster because seats are launched outside the CLI. Make the socket configurable and the roster written at launch, or make ps read tmux + seat dirs directly'
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: [I1]
+  - id: I3
+    title: Migrate fleet steering onto mosaic agent send --verify (exists, FLEET-OBS-005, spec FR-5) and retire tools/tmux/agent-send.sh, which forges the sender (D33) and returns an uninformative rc (D16, D34). FR-5 predates those defects by a month
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: [I1]
+  - id: I4
+    title: 'mosaic fleet absorbs what launch-seat.sh does and launch-seat.sh is deprecated: compose the prompt file set, force the skill set, wire the style hook, fail closed on any unreadable input, export per-seat git identity, and register the seat in the roster. launch-seat.sh was a manual method; it is the reference implementation, not the destination'
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: [I1, I5]
+  - id: I5
+    title: Harness probe matrix — verify a working prompt-injection path for claude, codex, opencode and pi, and refuse any runtime whose path is unverified. AC-NS-0 clause 1 ("any configured harness") rests on this. The probe work in docs/plans/2026-08-19_launch-seat-multi-runtime.md (brain, untracked) transfers; its launch-seat.sh target does not
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: []
+  - id: I6
+    title: Finish the heartbeat responder (FLEET-OBS-002, the only Phase-2 task still in-progress; spec FR-2). Health must mean "answered a heartbeat", not "pane alive" — pane state measured two seats wrong on 2026-08-20
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: []
+  - id: I7
+    title: Independent review and live-fleet dogfood of the Phase-2 verbs (FLEET-OBS-008), then land them (FLEET-OBS-009). Implementation is done and verification is not; "done" in a task file frozen five weeks is not evidence
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: [I2, I3, I4, I6]
+  - id: I8
+    title: Neutralize misleading documentation — supersede headers on docs that state a stale status, a false blocker or a retired mechanism. Cheap, and it is tier 0 because a stale doc does not merely fail to help an agent, it actively misroutes one. Rebuilding the documentation is a separate and later job
+    phase: 1
+    tier: 0
+    priority: must-have
+    depends_on: []
+  - id: I9
+    title: Study t3code's agent-attach and multi-provider auth methods and record what transfers. Reference only — Mosaic implements its own within the stack, never adopts the code and never takes the dependency. Informs HOW I/J/K/L are built, not whether
+    phase: 1
+    tier: 0
+    priority: should-have
+    depends_on: []
+  - id: J1
+    title: Web control plane over the gateway — fleet visibility and steering in a browser, same data source as I2
+    phase: 3
+    tier: 1
+    priority: must-have
+    depends_on: [I2]
+  - id: K1
+    title: Desktop and mobile clients against the gateway, authenticated
+    phase: 4
+    tier: 2
+    priority: must-have
+    depends_on: [J1]
+  - id: L1
+    title: Per-provider auth profiles with per-session selection
+    phase: 4
+    tier: 2
+    priority: must-have
+    depends_on: [I4]
 
 assumptions:
   - id: ASM-1

docs/fleet/PRD-fleet-suite.md

@@ -1,7 +1,7 @@
 # PRD — Mosaic Fleet Suite (init, configure, operate)
 
 > **Workstream:** W-FLEET (Fleet) under mission `mvp-20260312` · **Phase:** 3→4 productization
-> **North star:** [docs/fleet/north-star.md](./north-star.md) · prior: Phase-2 observability (#579), durable launch (#581), real-agent enablement (#583/#584/#586), releases 0.0.35–0.0.37
+> **North star:** [docs/fleet/FLEET-DOCTRINE.md](./FLEET-DOCTRINE.md) · prior: Phase-2 observability (#579), durable launch (#581), real-agent enablement (#583/#584/#586), releases 0.0.35–0.0.37
 > **Lead:** Jarvis @ `w-jarvis`. **Collaborator:** coder agent @ `dragon-lin` (jwoltje@10.1.10.37:coder0-0).
 > Owner of this file: Fleet workstream lead. Does not modify MVP single-writer control-plane files.
 

docs/fleet/PRD.md

@@ -1,7 +1,7 @@
 # PRD — Fleet Phase 2: Operator Observability
 
 > **Workstream:** W-FLEET under `mvp-20260312` · **Phase:** 2
-> **North star:** [docs/fleet/north-star.md](./north-star.md)
+> **North star:** [docs/fleet/FLEET-DOCTRINE.md](./FLEET-DOCTRINE.md)
 > **Source umbrella PRD:** [docs/PRD.md](../PRD.md) (Mosaic Stack v0.1.0)
 > **Tracks task:** `fleet-observability-1` — restore operator observability into fleet agent sessions.
 

docs/fleet/TASKS.md

@@ -1,10 +1,25 @@
 # Tasks — W-FLEET (Fleet) Phase 2: Observability
 
+> ---
+>
+> **STATUS: SUPERSEDED — 2026-08-20.** kind `tracking` · superseded by `docs/fleet/NORTH_STAR.yaml`
+>
+> This file is the pre-backlog tracking mechanism. `NS-2` in the north star declares the
+> replacement: every backlog item is a Mosaic Backlog card projected from the YAML. That
+> model replaced this one and nobody retired the old file, so it kept reading as
+> authoritative while going stale.
+>
+> **Do not trust a status in this file.** Verified 2026-08-20: it was already behind the
+> code when it froze five weeks ago. The `FLEET-OBS` series was the one thing worth salvaging and
+> is now carried as goals `I2`, `I3`, `I6` and `I7` at tier 0.
+>
+> Kept as a record of what was believed. Do not update it; update the YAML.
+
 > Workstream task file for the Fleet. Single-writer: Fleet workstream lead (orchestrator).
 > Workers read but never modify. This is **not** the MVP rollup (`docs/TASKS.md`) — a
 > rollup row is proposed to the MVP orchestrator, not written here.
 >
-> Mission: `mvp-20260312` · PRD: [docs/fleet/PRD.md](./PRD.md) · North star: [docs/fleet/north-star.md](./north-star.md)
+> Mission: `mvp-20260312` · PRD: [docs/fleet/PRD.md](./PRD.md) · North star: [docs/fleet/FLEET-DOCTRINE.md](./FLEET-DOCTRINE.md)
 > Status: `not-started` | `in-progress` | `done` | `blocked` | `failed`
 
 | id            | status      | description                                                                                                      | depends_on            | agent       | pr  | notes                                                                                                                       |

docs/fleet/f4-matrix-connector.md

@@ -1,6 +1,6 @@
 # F4 — Orchestrator chat connector + Matrix (local homeserver)
 
-> **Issue:** #616 · **Doctrine:** `docs/fleet/north-star.md` (#613) — orchestrator-chat-connector decision.
+> **Issue:** #616 · **Doctrine:** `docs/fleet/FLEET-DOCTRINE.md` (#613) — orchestrator-chat-connector decision.
 > **Status:** Phase 1 (abstraction + scaffold) in this PR; Phase 2+ are follow-ups (below).
 
 ## Goal

docs/native-kanban-sot/TASKS.md

@@ -4,7 +4,16 @@
 **Tracking:** [Mosaic Stack issue #751](https://git.mosaicstack.dev/mosaicstack/stack/issues/751)
 **Execution:** USC web1 only; collision-free GPT coder2/3/4/5 lanes
 **Contract:** `SHARED-CONTRACT.md` + four `contracts/*.v1.ts` files
-**Implementation hold:** no feature slice starts until the canon PR is merged to `main` with terminal-green CI; after merge, each slice remains held until every declared KBN prerequisite is complete.
+
+> **HOLD LIFTED — 2026-08-20.** The stated condition ("no feature slice starts until the
+> canon PR is merged to `main` with terminal-green CI") was **met on 2026-07-14** by
+> `49e8a541` — _docs(#751): Publish native Kanban/SOT canon (#752)_ — which is on both
+> `main` and `next`, and `docs/requirements/native-kanban-sot.md` is present on `main`.
+> The line below was never updated, so for five weeks this file advertised a blocker that
+> did not exist. The status line above ("PUBLICATION IN PROGRESS") is stale for the same
+> reason. This file is a build plan, not a task tracker, and is NOT superseded.
+
+**Implementation hold (SATISFIED — see note above):** no feature slice starts until the canon PR is merged to `main` with terminal-green CI; after merge, each slice remains held until every declared KBN prerequisite is complete.
 
 > This publication file is not a runtime task authority. After cutover, repository `TASKS.md` is generated read-only and never imported.
 

docs/release-integrity/TASKS.md

@@ -0,0 +1,42 @@
+# Tasks — Release Integrity Workstream (RI-050, #1275)
+
+> Single-writer: the RI-050 orchestrator (jarvis, dragon-lin) only. Workers read but never modify.
+>
+> **Mission:** alpha 0.0.50 release-integrity floor (decisions SDLC-D-033..038).
+> **PRD:** [docs/PRD.md § Release Integrity Workstream](../PRD.md#release-integrity-workstream-ri-1275)
+> **Issue:** #1275 (remains open until RI-V-001 closes)
+> **Base branch:** `next` (all cards branch from `origin/next`, squash-merge via PR)
+>
+> **Execution note:** the `agent` column uses `pi-glm-5.3` — outside the pipeline-cron model
+> table on purpose. This workstream is executed by jarvis on dragon-lin with local pi workers
+> (`pi --model zai/glm-5.3:high`); pipeline crons must not auto-claim these rows.
+>
+> **Status values:** `not-started` | `in-progress` | `done` | `blocked` | `failed` | `needs-qa`
+> `done` requires: repo quality gates green, independent review recorded, terminal-green CI on
+> the PR head, squash merge to `next`, and acceptance evidence in notes.
+
+| id       | status      | description                                                                                                                                                                                                                                                                      | issue | agent      | repo              | branch                            | depends_on                                                 | estimate | notes                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           |
+| -------- | ----------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----- | ---------- | ----------------- | --------------------------------- | ---------------------------------------------------------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| RI-0-001 | done        | Bootstrap: issue #1275, PRD section, this DAG, scratchpad (docs only)                                                                                                                                                                                                            | #1275 | pi-glm-5.3 | mosaicstack/stack | docs/ri-050-mission-bootstrap     | —                                                          | 6K       | PR #1276 (head 758659dd): docs-only, CI green (2475). Review requested from fargo. Merges first (no publish run).                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
+| RI-1-001 | done        | RI-N1: canonical terminal verification command + publish-pipeline exact-commit gate (every publish step depends on verify; commit identity check; fail closed)                                                                                                                   | #1275 | pi-glm-5.3 | mosaicstack/stack | feat/ri-050-publish-gate          | RI-0-001                                                   | 25K      | PR #1277 (head 46784c8d): CI GREEN at head after serialized retry (pipeline 2476, 2026-08-18) - earlier red was CI-agent contention (web SPA timeouts under concurrent pipelines), not code. Review requested from fargo at pinned head (comms 20260818T021025Z).                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
+| RI-1-002 | done        | RI-N1 negative control: checked-in tests proving a broken mandatory check blocks every publish step and that DAG edges cannot be bypassed                                                                                                                                        | #1275 | pi-glm-5.3 | mosaicstack/stack | test/ri-050-publish-gate-negative | RI-1-001                                                   | 12K      |                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| RI-2-001 | done        | RI-N2 (Forge): remove stub-executor false success; `--simulate` typed `simulated` results that satisfy nothing; literal-`true` gates and echo-review replaced with real gates or typed waiting-for-authority                                                                     | #1275 | pi-glm-5.3 | mosaicstack/stack | fix/ri-050-forge-fail-closed      | RI-0-001                                                   | 20K      | Independent review APPROVED 2026-08-17 (Gitea review 172 on PR #1278, head 99b8f6ea; reviewing seat fargo — recorded under shared host principal mos-dt-0, provenance correction posted by fred; wrapper gap filed by fred). Executed at head: forge tests 116/116, lint green, typecheck green after building macp dist (minimal-install artifact, not a defect), workspace typecheck 45/45, no external type consumers of the changed interfaces. CI red = known lane-wide fleet-test failure only, carries no information about this change (fred, log-content analysis, pipelines 2456-2458). Non-blocking finding: README L141-143 + skills/mosaic-forge/SKILL.md document bare forge run/resume, which now fails closed — fast-follow docs touch. Merge queued behind #1270. UPDATE 2026-08-18: #1270 merged; CI GREEN at head 4917df1f via serialized retry (pipeline 2477) - root cause of prior reds was CI-agent contention (web SPA timeouts under concurrent pipelines), superseding the fleet-test-failure theory. |
+| RI-2-002 | done        | RI-N2 (MACP): gate runner fails closed on empty commands, stub executors, and unimplemented CI-provider gates unless explicit simulate; typed capability failures                                                                                                                | #1275 | pi-glm-5.3 | mosaicstack/stack | fix/ri-050-macp-fail-closed       | RI-0-001                                                   | 15K      | PR #1293 (head 2097379e): CI green (pipeline 2465), independent review APPROVED (Gitea review 173, jarvis seat, 2026-08-17) - macp 109/109 verified at head. Merge queued behind #1276/#1277/#1278.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             |
+| RI-3-001 | done        | RI-N4: complete probe inventory mapping every TS and shell quality-rail check to one canonical check with disposition (preserve/strengthen/retire, each named)                                                                                                                   | #1275 | pi-glm-5.3 | mosaicstack/stack | docs/ri-050-qr-probe-inventory    | RI-0-001                                                   | 12K      | PR #1302 (head e06a47fac591): CI green (2484), independent review APPROVED (Gitea review 187, fargo seat, 2026-08-18) — 54 rows / 21 canonical checks / dispositions 43-2-9-0 verified by row-count and code spot-checks. Merged by fargo at pinned head.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       |
+| RI-3-002 | not-started | RI-N4: TS evaluator absorbs effective shell probes; typed results (passed/failed/blocked/error/not-applicable) with versioned digested check definitions; shell commands become thin adapters; contract/parity/negative-control tests                                            | #1275 | pi-glm-5.3 | mosaicstack/stack | feat/ri-050-qr-evaluator          | RI-3-001                                                   | 30K      |                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| RI-4-001 | in-progress | RI-N3: one PRD application service — `mission --plan` persists mission↔PRD linkage (ids/versions/selected requirements); `mosaic prdy` routes through the service or becomes a named import/export adapter; Markdown is a labeled generated view; explicit conflict-aware import | #1275 | pi-glm-5.3 | mosaicstack/stack | feat/ri-050-prd-authority         | RI-0-001                                                   | 35K      | PR #1294 (head 8d258e1d): CI green (pipeline 2466), independent review APPROVED (Gitea review 174, jarvis seat, 2026-08-17) - prdy 20/20 + command specs 9/9 at head. Merge queued behind #1276/#1277/#1278.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
+| RI-5-001 | done        | RI-N5: typed freshness states (current/stale/partial/unknown/unavailable); no failed-fetch-renders-empty; stale derived verdicts → unknown; mutations disabled when stale; failure-matrix tests                                                                                  | #1275 | pi-glm-5.3 | mosaicstack/stack | feat/ri-050-web-stale-safety      | RI-0-001                                                   | 25K      |                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| RI-V-001 | not-started | Final verification + release evidence: all cards verified merged, negative controls demonstrated, real `next` publish run green on exact commit, evidence pack recorded                                                                                                          | #1275 | pi-glm-5.3 | mosaicstack/stack | docs/ri-050-release-evidence      | RI-1-002, RI-2-001, RI-2-002, RI-3-002, RI-4-001, RI-5-001 | 10K      |                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+
+## Dispatch waves (max 2 parallel workers)
+
+1. RI-1-001 + RI-2-001
+2. RI-2-002 + RI-4-001
+3. RI-3-001 + RI-5-001
+4. RI-1-002 + RI-3-002
+5. RI-V-001
+
+## Budget
+
+Derived soft cap: 250K tokens (no explicit cap given). Projected total: 190K.
+Conservative mode (1 worker) above 70% projected; freeze above 90%.

docs/release-integrity/probe-inventory.md

@@ -0,0 +1,186 @@
+# Quality-Rails Probe Inventory — RI-3-001
+
+- **Task:** RI-3-001 (SDLC-D-037 first half; PRD § Release Integrity Workstream, RI-N4)
+- **Date:** 2026-08-18
+- **Base:** `origin/next` @ `8199261c` (branch `docs/ri-050-qr-probe-inventory`)
+- **Follow-up:** RI-3-002 consumes the dispositions here when building the single TS evaluator.
+
+## 0. Scope and method
+
+Every mechanism in this repository that verifies a quality, integrity, safety, or release
+property — TypeScript checks, shell probes, pipeline steps, git hooks, and installer-side
+assertions — gets one row. Each row's "what it actually verifies" was written from the
+probe's **code**, not its name or docs. Framework tool unit/regression suites (git wrappers,
+wake, tmux, orchestrator, …) are treated as one enforcement surface (`test:framework-shell`)
+because they test tool behavior rather than repo quality; their wiring integrity is itself
+guarded by `check-test-enumeration.sh`, and the quality-relevant members are rowed
+individually.
+
+**Kinds:** `ts` (TypeScript/Node check), `shell` (bash/python probe), `pipeline-step`
+(exists only inside a Woodpecker pipeline).
+
+**Enforcement points:** `local` (operator-invoked), `pre-commit`, `pre-push`,
+`CI ci.yml#<step>`, `publish.yml#<step>` (CI on push to main/next), `turbo <task>`,
+`agent-runtime` (framework hooks on an agent host), `installer` (host install path),
+`unwired`.
+
+**Dispositions** (recommendations for RI-3-002): `preserve` (keep as-is; already the
+canonical or a correct guard-of-the-guard), `strengthen` (keep, but a concrete gap must
+close — usually absorption into the TS evaluator), `strengthen (review)` (viable retirement
+candidate once the evaluator absorbs it; do not retire yet). Note: RI-N4 requires that
+effective shell probes be **absorbed before** their independent paths retire — no row here
+is marked `retire` because no absorption exists yet.
+
+## 1. Inventory
+
+### 1.1 Repo-level gate tasks (pnpm / turbo)
+
+| check                                 | location                                                                             | kind | what it actually verifies                                                                                                                                                                                                                                                                                                                                                                                                     | enforcement point                                                                                      | canonical check           | disposition | rationale                                                                                                                                                 |
+| ------------------------------------- | ------------------------------------------------------------------------------------ | ---- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | ------------------------- | ----------- | --------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `pnpm preflight` (checkout preflight) | `scripts/preflight.mjs`                                                              | ts   | Six gate binaries (eslint, husky, prettier, tsc, turbo, vitest) exist and are executable in `node_modules/.bin` (exit 42 if not); no stale `.mosaic-test-work/web-build.lock` (exit 43); `apps/web/.next` is a real directory (not a symlink), every entry owned by the current uid, and its `.mosaic-source-hash` fingerprint + `.mosaic-symlink-manifest` hash match the certified build written by `scripts/build-web.mjs` | `pre-push`; inside `pnpm typecheck` (→ `CI ci.yml#typecheck`, verify-release `typecheck` stage)        | QC-1 Checkout integrity   | preserve    | Blocks a poisoned/stale generated `.next` from faking a green typecheck (the five-month-stale-`.next` class); trust chain is self-contained per-checkout. |
+| `pnpm typecheck`                      | root `package.json` → `turbo run typecheck`                                          | ts   | Per-package `tsc --noEmit` (all 20 packages); turbo `typecheck` depends on `^build`, so package builds must succeed first; prefixed by checkout preflight                                                                                                                                                                                                                                                                     | `CI ci.yml#typecheck`; `pre-push`; verify-release `typecheck` stage; `turbo typecheck`                 | QC-2 Workspace typecheck  | preserve    | The single workspace-wide type gate; CI and hooks invoke the same task, no divergent checklist.                                                           |
+| `pnpm lint`                           | root `package.json` → `turbo run lint`                                               | ts   | Per-package `eslint src` under root `eslint.config.mjs` (ignores `dist`, `.next`, `framework/**`, etc.)                                                                                                                                                                                                                                                                                                                       | `CI ci.yml#lint`; `pre-push`; verify-release `lint` stage; `turbo lint`                                | QC-3 Workspace lint       | preserve    | Same-task invocation from every surface; no second lint definition.                                                                                       |
+| `pnpm format:check`                   | root `package.json` → `prettier --check`                                             | ts   | Prettier parse/format equality over `**/*.{ts,tsx,js,jsx,json,md}` minus `.prettierignore` (generated trees, `docs/scratchpads/`, venvs, …)                                                                                                                                                                                                                                                                                   | `CI ci.yml#format`; `pre-push`; verify-release `format` stage                                          | QC-4 Format check         | preserve    | Single formatter, single ignore list, enforced identically everywhere.                                                                                    |
+| `pnpm test`                           | root `package.json` `test` = `test:checkout` && `turbo run test` && `test:installer` | ts   | (a) `node --test scripts/*.test.mjs` — checkout-tool units; (b) per-package `vitest run` (mosaic appends the 47-command `test:framework-shell` chain); (c) `tools/install-next-lane.test.sh`; turbo `test` declares DB env vars and depends on `^build`                                                                                                                                                                       | `CI ci.yml#test` (with `DATABASE_URL` + `db:migrate` first); verify-release `test` stage; `turbo test` | QC-5 Test suite execution | preserve    | One composed test command; the chain property (any link red ⇒ step red) is the gate.                                                                      |
+| `pnpm build`                          | root `package.json` → `turbo run build`                                              | ts   | Per-package build (`tsc`/Next) with `^build` dependency and `dist/**` outputs                                                                                                                                                                                                                                                                                                                                                 | `publish.yml#build`; verify-release `build` stage; `turbo build`                                       | QC-6 Workspace build      | preserve    | Publish artifacts derive from the same build task CI verifies.                                                                                            |
+
+### 1.2 Framework quality shell probes (`packages/mosaic/framework/tools/quality/`)
+
+| check                                       | location                                                                | kind  | what it actually verifies                                                                                                                                                                                                                                                                                                                                                                                                                                                                        | enforcement point                                                                                                                                                                                                                                                                                                                                                           | canonical check                                                                                                                         | disposition                      | rationale                                                                                                                                                                                                                                                         |
+| ------------------------------------------- | ----------------------------------------------------------------------- | ----- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- |
+| Sanitization gate                           | `scripts/verify-sanitized.sh`                                           | shell | Built-in self-test first (planted identity/structural/YAML+service fixtures; exit 2 if the regexes or extension coverage break), then: (1) identity denylist grep (`jarvis\|jason\|woltje\|brain.woltje.com\|/home/jwoltje\|\bPDA\b`) over all shipped text files **including** `examples/`; (2) structural grep for private `$HOME/src` defaults in shipped scripts **excluding** `examples/`. Any hit ⇒ exit 1                                                                                 | `CI ci.yml#sanitization`; verify-release `sanitization` stage                                                                                                                                                                                                                                                                                                               | QC-7 Framework sanitization                                                                                                             | preserve                         | Labeled one-time regression guard with a self-test that prevents silent no-op; correctly scoped (identity vs structural) and documented as not a general PII detector.                                                                                            |
+| Resident-context budget                     | `scripts/check-resident-budget.sh` (+ `--self-test`)                    | shell | Self-test of the comparator, then `wc -l` vs per-file ceilings (CONSTITUTION 120, AGENTS 120, each RUNTIME.md 90); missing file ⇒ fail; over ceiling ⇒ exit 1                                                                                                                                                                                                                                                                                                                                    | `CI ci.yml#sanitization` (both modes); verify-release `sanitization` stage                                                                                                                                                                                                                                                                                                  | QC-8 Resident-context budget                                                                                                            | preserve                         | Caps the container (lines), never the wording — the deliberate anti-drift design (DESIGN §7); CI-enforceable half only, by design.                                                                                                                                |
+| Test-membership enumeration guard (#1017)   | `scripts/check-test-enumeration.sh` + `test-enumeration-exclusions.txt` | shell | Parses surface S1 (`packages/mosaic` `test:framework-shell` via JSON+shlex) and S2 (every `framework/tools/\*.sh                                                                                                                                                                                                                                                                                                                                                                                 | .py`token in`ci.yml`, comment lines stripped); population = `_test_.sh`under`framework/tools`; FAILS on: suite-shaped file on disk neither enumerated nor signed-excluded; surface naming a path missing on disk (both directions); exclusion without reason / stale / outside population / contradicting enumeration. Proves **naming, not reachability** (stated in-file) | `CI ci.yml#sanitization` (direct line); link [0] of `test:framework-shell` (thus `CI ci.yml#test`); verify-release `sanitization` stage | QC-9 Test-membership enumeration | preserve                                                                                                                                                                                                                                                          | Makes silent under-run impossible; invoked from both surfaces it audits so severing the chain cannot silence it. |
+| Enumeration-guard needles                   | `scripts/test-check-test-enumeration.sh`                                | shell | Needle/control fixtures driven through `--root`: every promised failure mode must trip the guard **on its own words**, plus controls that must pass (null-case defense); covers commented-out ci.yml lines (F1) and line-range parsing (n2b)                                                                                                                                                                                                                                                     | `test:framework-shell` → `CI ci.yml#test`; verify-release `test` stage                                                                                                                                                                                                                                                                                                      | QC-9 Test-membership enumeration                                                                                                        | preserve                         | Guard-of-the-guard with both polarities; same canonical check by design.                                                                                                                                                                                          |
+| Upgrade manifest guard (#791 HARD GATE)     | `scripts/test-upgrade-manifest-guard.sh`                                | shell | Keep-mode `install.sh` upgrade against seeded throwaway `MOSAIC_HOME`: every operator sentinel — including an **unanticipated** one — survives byte-identical with unchanged mtime; framework files still update; retired framework files pruned; matrix run with rsync present AND absent (keep path must be rsync-independent); fail-closed matrix (empty/operator-only/malformed/missing manifest aborts loudly, operator files untouched); operator secret never appears in installer output | `CI ci.yml#upgrade-guard`; verify-release `upgrade-guard` stage                                                                                                                                                                                                                                                                                                             | QC-10 Upgrade/install safety                                                                                                            | preserve                         | The operator-data hard gate for the `mosaic update` path; negative controls are load-bearing and documented.                                                                                                                                                      |
+| Upgrade rollback gate (#791 B1)             | `scripts/test-upgrade-rollback.sh`                                      | shell | Mid-sync failure (PATH-shadowing `cp` shim) must trigger snapshot restore: restore message fires, corrupted file restored, target byte-identical to pre-upgrade; control installer with `set -E` stripped must NOT roll back (proves errtrace is load-bearing); plus signal/exit-guard controls                                                                                                                                                                                                  | `CI ci.yml#upgrade-guard`; verify-release `upgrade-guard` stage                                                                                                                                                                                                                                                                                                             | QC-10 Upgrade/install safety                                                                                                            | preserve                         | Proves the rollback trap actually fires; the `-E`-stripped control keeps Part A honest.                                                                                                                                                                           |
+| Durable-snapshot gate (#791 PR2)            | `scripts/test-upgrade-durable-snapshot.sh`                              | shell | Pre-update snapshot taken before any mutation (0700/0600 perms, secret never logged, retention-pruned); post-sync verify net restores operator files a manifest bug lets the sync touch; CWE-59 symlink-leaf guard proven with a portable cp shim in both polarities (write-through-link must not happen); v1→v2 migration semantics (intended `bin/` removal not healed)                                                                                                                        | `CI ci.yml#upgrade-guard`; verify-release `upgrade-guard` stage                                                                                                                                                                                                                                                                                                             | QC-10 Upgrade/install safety                                                                                                            | preserve                         | Covers tampering and leak vectors the manifest guard cannot see; the shim rationale (busybox vs GNU cp) is documented in-file.                                                                                                                                    |
+| Install migration matrix (v2→v3)            | `scripts/test-install-migration.sh`                                     | shell | Fixture matrix running the real installer with `MOSAIC_SYNC_ONLY=1`: fresh install seeds + stamps version 3; legacy user-edited AGENTS overwritten with `.pre-constitution.bak` preserved (and idempotent); tuned STANDARDS overwritten; operator files (SOUL, credentials) preserved. Mirrors the TS suite `packages/mosaic/src/config/file-adapter.test.ts` — both installers must behave identically                                                                                          | `CI ci.yml#upgrade-guard`; verify-release `upgrade-guard` stage                                                                                                                                                                                                                                                                                                             | QC-10 Upgrade/install safety                                                                                                            | preserve                         | Pins the shell/TS installer parity contract; removal would orphan that parity requirement.                                                                                                                                                                        |
+| Enforcement verification probe (bash)       | `scripts/verify.sh`                                                     | shell | Attempts **real commits** in the target repo: planted type error must produce a commit blocked with `error`; planted `any` must trip `no-explicit-any`; planted lint error must trip `prettier`; gitleaks binary must exist (3a) and detect a planted AWS key via `gitleaks git --pre-commit --staged --redact` (3b). Verdicts are output-grep matches on hook stderr                                                                                                                            | `local` via installed `mosaic-quality-verify` on scaffolded target projects; **not run in this repo's CI**                                                                                                                                                                                                                                                                  | QC-20 Downstream enforcement verification                                                                                               | strengthen (review)              | Mechanism is genuinely behavioral (stronger than file presence) but verdict logic is grep-on-output and it is unwired here; absorb as the evaluator's enforcement-probe check (the RI-N4 evaluator invokes it or reimplements it) before retiring the shell path. |
+| Enforcement verification probe (PowerShell) | `scripts/verify.ps1`                                                    | shell | Windows port of `verify.sh`: same planted-commit tests with `$output -match` matching; no gitleaks self-test parity beyond the same checks                                                                                                                                                                                                                                                                                                                                                       | `local` (Windows operator); no Windows CI runner exists                                                                                                                                                                                                                                                                                                                     | QC-20 Downstream enforcement verification                                                                                               | strengthen (review)              | A hand-maintained twin of `verify.sh` with no CI coverage — exactly the drift shape the single evaluator removes; retire after the TS evaluator owns the probe.                                                                                                   |
+| Quality template installer (bash)           | `scripts/install.sh`                                                    | shell | Copies template files (`.husky/pre-commit` incl. mandatory gitleaks, `.lintstagedrc.js`, `.eslintrc.js`, `tsconfig.json`, `.woodpecker.yml`, `.gitleaks.toml`) into a target project; **warns** (does not verify) about `package.json` snippet merge; no post-condition check                                                                                                                                                                                                                    | `local` / via `mosaic-quality-apply`                                                                                                                                                                                                                                                                                                                                        | QC-21 Downstream rails scaffolding                                                                                                      | strengthen (review)              | Duplicates the TS `quality-rails init` scaffolder for a different template set; converging on one scaffolder (with post-scaffold verification) is prerequisite to retiring this path.                                                                             |
+| Quality template installer (PowerShell)     | `scripts/install.ps1`                                                   | shell | Windows twin of the template copy above                                                                                                                                                                                                                                                                                                                                                                                                                                                          | `local` (Windows operator)                                                                                                                                                                                                                                                                                                                                                  | QC-21 Downstream rails scaffolding                                                                                                      | strengthen (review)              | Same twin-drift risk as `verify.ps1`; no runner exercises it.                                                                                                                                                                                                     |
+| `mosaic-quality-verify` adapter             | `framework/tools/_scripts/mosaic-quality-verify`                        | shell | Thin adapter: validates target dir exists, asserts `verify.sh` present+executable, `cd` target, exec it. No verdict logic of its own                                                                                                                                                                                                                                                                                                                                                             | `local` (installed framework bin)                                                                                                                                                                                                                                                                                                                                           | QC-20 Downstream enforcement verification                                                                                               | preserve                         | Already the thin-adapter shape RI-N4 prescribes for shell surfaces.                                                                                                                                                                                               |
+| `mosaic-quality-apply` adapter              | `framework/tools/_scripts/mosaic-quality-apply`                         | shell | Thin adapter: arg validation then exec of quality `install.sh --template … --target …`                                                                                                                                                                                                                                                                                                                                                                                                           | `local` (installed framework bin)                                                                                                                                                                                                                                                                                                                                           | QC-21 Downstream rails scaffolding                                                                                                      | preserve                         | Thin adapter, no separate verdict; disposition follows its target script's convergence.                                                                                                                                                                           |
+| Roster schema regression                    | `scripts/test-roster-schema.py`                                         | shell | jsonschema `Draft202012Validator` over `fleet/roster.schema.json` with valid/invalid connector-kind fixtures (tmux/discord/matrix conditional fields)                                                                                                                                                                                                                                                                                                                                            | **unwired** — not on S1 or S2, not signed-excluded; also outside the enumeration guard's `*.sh` population, so the guard cannot see it                                                                                                                                                                                                                                      | QC-5 Test suite execution                                                                                                               | strengthen (review)              | A real regression suite that currently runs nowhere; wire it into a CI surface or sign an exclusion — leaving it invisible re-arms the exact gap #1017 closed.                                                                                                    |
+| Framework shell chain (S1)                  | `packages/mosaic/package.json` `test:framework-shell`                   | shell | 47-command `&&` chain: enumeration guard + needles, 14 lease-broker/mutator-gate python unitests, `check-runtime-launches.py`, and ~30 framework-tool shell suites (git wrappers, wake, woodpecker, tmux, glpi, orchestrator, `_scripts`). Quality-relevant members rowed separately below                                                                                                                                                                                                       | `turbo test` → `CI ci.yml#test`; verify-release `test` stage                                                                                                                                                                                                                                                                                                                | QC-5 Test suite execution                                                                                                               | preserve                         | The chain is the execution surface the enumeration guard audits; known residuals: a failing link stops later suites (measured in #1270 — suites after position 44 had not run), and the guard proves naming, not reachability.                                    |
+
+### 1.3 Framework runtime hooks and their harnesses (agent-host enforcement)
+
+| check                                 | location                                                                                  | kind  | what it actually verifies                                                                                                                                                                                                                                                                                                                                                            | enforcement point                                                                | canonical check                         | disposition         | rationale                                                                                                                                                                                                  |
+| ------------------------------------- | ----------------------------------------------------------------------------------------- | ----- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------- | --------------------------------------- | ------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| QA edit hook seam                     | `framework/tools/qa/qa-hook-stdin.sh` (+ `qa-hook-handler.sh`)                            | shell | PostToolUse stdin hook: extracts edited file from the tool JSON (jq or grep fallback), skips non-JS/TS, then the deps-preflight gate — exits 1 with the legible sentinel `deps not installed — run pnpm install` when `node_modules/.bin` is missing/empty (the #856 false-red class); the downstream handler only files QA remediation **report templates** (no verification logic) | `agent-runtime` (framework `runtime/claude/settings.json` PostToolUse); never CI | QC-16 Agent-runtime edit-time checks    | strengthen (review) | The sentinel gate is real enforcement; the handler's report-filing adds no verdict and its name promises more than the code does — evaluator absorption should keep the sentinel, drop the report theater. |
+| Typecheck-on-edit hook                | `framework/tools/qa/typecheck-hook.sh`                                                    | shell | PostToolUse: for edited `.ts/.tsx`, finds nearest `tsconfig.json` and runs `tsc --noEmit`, surfacing errors nonzero to the agent immediately                                                                                                                                                                                                                                         | `agent-runtime` (framework `runtime/claude/settings.json` PostToolUse)           | QC-16 Agent-runtime edit-time checks    | strengthen (review) | Edit-time duplicate of QC-2 with independent invocation logic; keep behavior, converge invocation through the evaluator adapter.                                                                           |
+| Deps-preflight harness                | `framework/tools/qa/test-deps-preflight.sh`                                               | shell | Five assertions against the seam incl. a documented RED control (raw `not found`), sentinel behavior for missing and empty `.bin`, and no-false-positive once populated                                                                                                                                                                                                              | `test:framework-shell` → `CI ci.yml#test`                                        | QC-16 Agent-runtime edit-time checks    | preserve            | Guard-of-the-check with a red control; keeps the sentinel from regressing.                                                                                                                                 |
+| Prompt-helper RCE regression          | `framework/tools/_scripts/test-mosaic-init-rce.sh`                                        | shell | Sources the prompt helpers and proves a literal `$(touch /tmp/pwned)` answer round-trips verbatim and never executes (no `/tmp/pwned` created)                                                                                                                                                                                                                                       | `test:framework-shell` → `CI ci.yml#test`                                        | QC-5 Test suite execution               | preserve            | Cheap, load-bearing security regression on the installer's input path.                                                                                                                                     |
+| Install-ordering harness (#869 C2)    | `framework/tools/_scripts/test-install-ordering-guard.sh`                                 | shell | Drives `mosaic-link-runtime-assets` with a fake `mosaic` on PATH: probe ok ⇒ settings copied + exit 0; probe fail ⇒ exit 1 with degraded outcome but all other runtime files still copied; `--allow-inactive-enforcement` forwarded; no-mosaic-on-PATH ⇒ python3 fallback strips enforcement hooks and exits 1; fallback + flag ⇒ wires as-is, exit 0                                | `test:framework-shell` → `CI ci.yml#test`                                        | QC-17 Lease-enforcement wiring safety   | preserve            | Exercises the shell wiring seam independently of the TS guard's own spec suite (complementary coverage, by design).                                                                                        |
+| Fleet-transport harness (#1240)       | `framework/tools/_scripts/test-fleet-transport-check.sh`                                  | shell | Extracts the shipped `check_fleet_transport`/`fleet_declared_transport` functions **from the shipped scripts** (fails loud if extraction yields nothing) and drives both implementations (mosaic-doctor + `tools/install.sh`) from one case table                                                                                                                                    | `test:framework-shell` → `CI ci.yml#test`                                        | QC-18 Operator-host drift audit         | preserve            | The anti-drift harness for the one rule shipped twice; extraction-from-source keeps it from testing a stale copy.                                                                                          |
+| Terminal-green contract (RM-61/#1000) | `framework/tools/woodpecker/test-terminal-green-contract.sh` + `verify-terminal-green.py` | shell | Red-first fixtures: pipeline JSON variants (service failure, step failure, cancelled, etc.) must produce the correct terminal-green verdict; controls must pass                                                                                                                                                                                                                      | `test:framework-shell` → `CI ci.yml#test`                                        | QC-5 Test suite execution               | preserve            | Keeps the CI-wait wrapper's green-detection honest; a false green here would poison every merge gate that trusts `pr-ci-wait.sh`.                                                                          |
+| Lease-gate launch invariant           | `framework/tools/lease-broker/check-runtime-launches.py`                                  | shell | Scans production roots (`packages/`, `apps/`, `plugins/`, `tools/`) across sh/py/ts/yaml suffixes for Claude/Pi process launches **outside** the lease gate; allowlist-based; fails CI on violation                                                                                                                                                                                  | `test:framework-shell` → `CI ci.yml#test`                                        | QC-15 Lease-gate architecture invariant | preserve            | The only architectural "no ungated launches" rail; grep+allowlist is the right cost/benefit for this invariant.                                                                                            |
+
+### 1.4 TypeScript quality logic (`@mosaicstack/quality-rails` + mosaic CLI)
+
+| check                                    | location                                                                                       | kind     | what it actually verifies                                                                                                                                                                                                                                                                                                                                                                                                            | enforcement point                                                                                              | canonical check                       | disposition         | rationale                                                                                                                                                                                                                                                                           |
+| ---------------------------------------- | ---------------------------------------------------------------------------------------------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------- | ------------------------------------- | ------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `quality-rails check`                    | `packages/quality-rails/src/cli.ts` (`mosaic quality-rails check --project`)                   | ts       | **Expected-file presence only**: loops `expectedFilesForKind` (node: `.eslintrc`, `biome.json`, `.githooks/pre-commit`, `PR-CHECKLIST.md`; python: `pyproject.toml`+hooks+checklist; rust: `rustfmt.toml`+…) and exits 1 listing missing paths. Does not execute any linter, formatter, hook, or scanner                                                                                                                             | `local` (operator CLI); **no CI wiring in this repo**                                                          | QC-19 Downstream rails presence check | strengthen          | This is the RI-N4 evaluator seed. Today presence ≠ parity (explicitly called out by RI-N4): it must grow typed verdicts (`passed/failed/blocked/error/not-applicable`), check versioning/subject/reason, digested definitions, and absorb the effective shell probes (QC-20 first). |
+| `quality-rails doctor`                   | `packages/quality-rails/src/cli.ts`                                                            | ts       | Same presence data as `check`, printed with ok/missing lines; **cannot fail** (no nonzero exit on missing files)                                                                                                                                                                                                                                                                                                                     | `local` (operator CLI)                                                                                         | QC-19 Downstream rails presence check | strengthen          | A doctor that cannot fail is advisory; fold into `check` (or return typed states) when the evaluator lands.                                                                                                                                                                         |
+| `quality-rails init`                     | `packages/quality-rails/src/cli.ts` + `scaffolder.ts`/`templates.ts`                           | ts       | Scaffolds rails files per detected kind/profile (linters/formatters lists are advisory strings; hooks flag always true); writes files, prints follow-ups — no post-condition verification                                                                                                                                                                                                                                            | `local` (operator CLI)                                                                                         | QC-21 Downstream rails scaffolding    | strengthen (review) | Second scaffolding path alongside quality `install.sh` (§1.2); converge on one with post-scaffold verification before retiring either.                                                                                                                                              |
+| Lease activation probe (#869 C1, hidden) | `packages/mosaic/src/commands/lease-activation-probe.ts`                                       | ts       | Real capability probe, not file presence: resolves the installed mosaic CLI and requires it to advertise the exact `{name, version}` activation contract; all deps injectable; registered as hidden CLI command and consumed by C2/C5                                                                                                                                                                                                | `local` (hidden CLI + consumed by C2/C5); spec-tested via `lease-activation-probe.spec.ts` in `turbo test`     | QC-17 Lease-enforcement wiring safety | preserve            | The versioned-contract probe is precisely the fail-closed capability check RI-N2 generalizes; already typed and injectable.                                                                                                                                                         |
+| Install-ordering guard (#869 C2, hidden) | `packages/mosaic/src/commands/install-ordering-guard.ts`                                       | ts       | Decides whether enforcement hook entries are written into the `~/.claude/settings.json` the framework reseed ships: not activatable ⇒ strip hooks + nonzero loud outcome (default); explicit per-invocation `--allow-inactive-enforcement` opt-out wires-with-warning. Never touches the runtime gate's own fail-closed behavior                                                                                                     | `installer` (framework reseed via `mosaic-link-runtime-assets`); spec + shell harness coverage in `turbo test` | QC-17 Lease-enforcement wiring safety | preserve            | Correct default-deny with an explicit, non-env opt-out; test-locked from both the TS and shell sides.                                                                                                                                                                               |
+| Lease doctor check (#869 C5)             | `packages/mosaic/src/commands/lease-doctor-check.ts`                                           | ts       | Combines hook-wiring detection in `~/.claude/settings.json` with C1 activatable and C3 broker-supervisor health: wired ∧ (¬activatable ∨ ¬healthy) ⇒ loud `[ERROR]` that forces `mosaic doctor` exit 1 regardless of the bash audit's own exit                                                                                                                                                                                       | `local` (inside `mosaic doctor`); spec coverage in `turbo test`                                                | QC-17 Lease-enforcement wiring safety | preserve            | Closes the "bricked host looks green" hole; cannot be masked by the bash script — that composition is the point.                                                                                                                                                                    |
+| `mosaic doctor` (framework drift audit)  | `packages/mosaic/src/commands/launch.ts` (`doctor`) + `framework/tools/_scripts/mosaic-doctor` | shell+ts | Bash audit of the installed framework home: ~40 expected files/dirs present; runtime files are copies (not symlinks) matching source (`cmp`) or composed runtime-contract markers; hard-gates block present in AGENTS.md; sequential-thinking MCP configured; fleet transport binary present per roster (warn); legacy symlink trees gone; skills synced — **warn-based, exit 1 only with `--fail-on-warn`**, plus C5's forced error | `local` (operator audit)                                                                                       | QC-18 Operator-host drift audit       | preserve            | Host-state audit CI cannot see (user files by design, DESIGN §7); advisory exit is the documented contract — do not silently change it.                                                                                                                                             |
+| `mosaic gateway doctor`                  | `packages/mosaic/src/commands/gateway-doctor.ts`                                               | ts       | Probes per-service health (PostgreSQL, Valkey, pgvector) via `@mosaicstack/storage`, reports tier and JSON; exit 1 only when at least one **required** service fails (yellow stays 0)                                                                                                                                                                                                                                                | `local` (operator)                                                                                             | QC-18 Operator-host drift audit       | preserve            | Service health with correct red/yellow exit semantics; JSON mode exists for scripting.                                                                                                                                                                                              |
+| `mosaic gateway verify`                  | `packages/mosaic/src/commands/gateway/verify.ts`                                               | ts       | Post-install liveness: daemon meta via HTTP with retries, admin token on file, bootstrap endpoint reachable; aggregated pass/fail                                                                                                                                                                                                                                                                                                    | `local`; consumed by `tools/e2e-install-test.sh`                                                               | QC-18 Operator-host drift audit       | preserve            | The first-run proof the installer E2E relies on; retry-aware so startup races don't false-red.                                                                                                                                                                                      |
+| `mosaic fleet doctor`                    | `packages/mosaic/src/commands/fleet-reconciler-command.ts`                                     | ts       | Classifies local roster-owned drift (no mutation) from the parsed v2 roster                                                                                                                                                                                                                                                                                                                                                          | `local` (operator)                                                                                             | QC-18 Operator-host drift audit       | preserve            | Dry-run classification is the correct non-mutating audit shape.                                                                                                                                                                                                                     |
+
+### 1.5 Git hooks (developer machine)
+
+| check                     | location                                                  | kind  | what it actually verifies                                                                                                                                                                                                                                | enforcement point                                                                                       | canonical check             | disposition | rationale                                                                                                         |
+| ------------------------- | --------------------------------------------------------- | ----- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------- | --------------------------- | ----------- | ----------------------------------------------------------------------------------------------------------------- |
+| Pre-commit staged hygiene | `.husky/pre-commit` → `npx lint-staged` (`.lintstagedrc`) | shell | On staged files only: `prettier --write` + `eslint --fix` for ts/tsx/js/jsx; `prettier --write` for json/md/yaml/yml. **Mutating** (fixes and re-stages); commit blocks only if a fixer itself fails                                                     | `pre-commit` (every local commit; hooks activated by `install-hooks.mjs` via `core.hooksPath .husky/_`) | QC-13 Staged-change hygiene | preserve    | Correct scoped fast gate; note it auto-fixes rather than rejects (deliberate). Gap: no secret scan here — see §3. |
+| Pre-push gate             | `.husky/pre-push`                                         | shell | `pnpm preflight && pnpm typecheck && pnpm lint && pnpm format:check` (no test run — documented in AGENTS.md)                                                                                                                                             | `pre-push`                                                                                              | QC-14 Pre-push gate         | preserve    | Composes QC-1..4 exactly as specified in AGENTS.md; tests intentionally left to CI.                               |
+| Hook installer            | `scripts/install-hooks.mjs` (`pnpm prepare`)              | ts    | Stages husky hooks into a scratch repo first, asserts husky produced its `h` shim, quarantines incomplete previous sets, verifies idempotence via full directory snapshot comparison, then sets `core.hooksPath`; skips cleanly with `HUSKY=0` or no git | `installer` (runs on `pnpm install`)                                                                    | QC-13 Staged-change hygiene | preserve    | Self-verifying wiring for the hook gates — a corrupted half-install cannot silently disable them.                 |
+
+### 1.6 CI pipeline steps (`.woodpecker/`)
+
+Step-to-probe mapping for container steps: `ci.yml#sanitization` = QC-7+QC-8+QC-9 (rows §1.2, plus `apk add bash` env prep); `ci.yml#upgrade-guard` = QC-10 (rows §1.2, plus `apk add rsync`); `ci.yml#typecheck`/`#lint`/`#format`/`#test` = QC-2/3/4/5 (rows §1.1). Rows below are mechanisms that exist only in a pipeline.
+
+| check                                  | location                                                                               | kind          | what it actually verifies                                                                                                                                                                                                                                                                                                            | enforcement point                                    | canonical check                     | disposition                                                           | rationale                                                                                                     |
+| -------------------------------------- | -------------------------------------------------------------------------------------- | ------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------- | ----------------------------------- | --------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------- | -------- | ------------------------------------------------------------------------------------------------------------- |
+| Frozen install                         | `ci.yml#install`                                                                       | pipeline-step | `pnpm install --frozen-lockfile --prefer-offline` against the baked ci-base store — lockfile supply integrity; a drifted lockfile fails the build before any gate runs                                                                                                                                                               | `CI ci.yml#install`                                  | QC-1 Checkout integrity             | preserve                                                              | Lockfile-pinned dep resolution is the supply-chain floor under every later gate.                              |
+| Test-step readiness prelude            | `ci.yml#test` prologue                                                                 | pipeline-step | Installs pinned `@earendil-works/[email protected]` (Invariant R suite requires the real binary) + openssl; waits up to 60×1s on `pg_isready` for the `ci-postgres` service and fails fast if it never comes up; runs `db:migrate` before tests                                                                                 | `CI ci.yml#test`                                     | QC-5 Test suite execution           | preserve                                                              | Fail-fast environment preconditions — a missing service produces a legible failure, not a wall of red tests.  |
+| Publish verify step (pending RI-1-001) | `publish.yml#verify` (branch `feat/ri-050-publish-gate` @ `46784c8d`, not yet on next) | pipeline-step | (a) Commit identity: fails closed if `CI_COMMIT_SHA` empty, `git rev-parse HEAD` empty, or the two differ; (b) runs the canonical `pnpm verify:release`. **Every publish effect depends on this step; it carries no path filter**                                                                                                    | `publish.yml#verify`                                 | QC-11 Terminal release verification | preserve                                                              | The RI-N1 exact-commit binding; until it merges, publish steps on next depend on `build` only (see §3 gap 1). |
+| Publish error classification           | `publish.yml#publish-npm`                                                              | pipeline-step | Publishes `@mosaicstack/*` (minus web) and classifies outcome: success, or the **only tolerated failure** = already-published (EPUBLISHCONFLICT / "cannot publish over" / "previously published"); explicit fatal on npm `E404/E401/ENEEDAUTH/ECONNREFUSED/ETIMEDOUT/ENOTFOUND` and on any unrecognized failure (replacing the old ` |                                                      | echo` that hid a registry 404)      | `publish.yml#publish-npm` (main/tags, path-filtered on `packages/**`) | QC-12 Publish-effect integrity                                                                                | preserve | Converts silent publish fall-on-floor into loud failure; allowlist-of-one error tolerance is the right shape. |
+| Next-lane publish assertions           | `publish.yml#publish-next-npm`                                                         | pipeline-step | Guards: branch must be `next`, `CI_PIPELINE_NUMBER` required; registry dist-tags JSON must be usable; walks all manifests, strictly parses stable semver, rewrites `X.Y.(Z+1)-next.<N>`; publishes with `--tag next` (never latest); post-publish asserts `npm view @mosaicstack/mosaic@next` resolves to the exact expected version | `publish.yml#publish-next-npm` (push/manual on next) | QC-12 Publish-effect integrity      | preserve                                                              | Durable prerelease lane with end-to-end resolution proof — the published artifact is verified, not assumed.   |
+| Image destination policy               | `publish.yml#build-gateway` / `#build-appservice` / `#build-web`                       | pipeline-step | Kaniko builds with destination policy: `next` ⇒ sha-tag only (fatal if a tag event sneaks in); `main` ⇒ sha + `latest`; tag events ⇒ sha + `<tag>`; anything else fatal. Path filters only skip **effects**, never the verify step                                                                                                   | `publish.yml#build-*`                                | QC-12 Publish-effect integrity      | preserve                                                              | Fail-closed tagging matrix; the exclude-list default-safe design keeps stale images impossible.               |
+
+Adjacent pipeline surface (not a probe): `.woodpecker/ci-image.yml` rebuilds the ci-base image on `pnpm-lock.yaml`/`Dockerfile.ci` change with an immutable `lock-<hash>` tag; pipelines consume `:latest`. Recorded for completeness — no code-quality property is checked.
+
+### 1.7 Root installer tooling (`tools/`)
+
+| check                       | location                                                  | kind  | what it actually verifies                                                                                                                                                                                                                                                              | enforcement point                                       | canonical check                 | disposition         | rationale                                                                                                                                                       |
+| --------------------------- | --------------------------------------------------------- | ----- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------- | ------------------------------- | ------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| Next-lane installer test    | `tools/install-next-lane.test.sh` (`pnpm test:installer`) | shell | Drives `tools/install.sh --next` with faked `node`/`npm` binaries (no network): Node 20 must be rejected; installs must pin **exact** versions (mutable `@next` forbidden); fast path must not unexpectedly fall back to source; gateway-install failure takes the documented fallback | `turbo`-external tail of `pnpm test` → `CI ci.yml#test` | QC-5 Test suite execution       | preserve            | Hermetic (shimmed) regression net for the installer lane; runs as part of the standard test command.                                                            |
+| Clean-container install E2E | `tools/e2e-install-test.sh`                               | shell | Full first-run flow in a node:22-alpine container: `install.sh --yes` → `mosaic wizard` (non-interactive) → `mosaic gateway install` → `mosaic gateway verify` exit check (with EXPECTED-SKIP if the installed CLI predates `gateway verify`); skips gracefully without Docker         | `local` (manual; requires Docker); **not wired in CI**  | QC-5 Test suite execution       | strengthen (review) | The only end-to-end proof of the install→verify path; currently operator-initiated only — wire into a periodic/manual CI lane or sign its exclusion explicitly. |
+| Host installer advisories   | `tools/install.sh` (`--check`; `check_fleet_transport`)   | shell | `--check` = version comparison only, no install; `check_fleet_transport` warns (non-blocking, by design — tmux is the fleet's dependency, not mosaic's) when the roster-declared transport binary is absent, naming exactly what it blocks; PATH-persistence warnings                  | `installer` (operator-run)                              | QC-18 Operator-host drift audit | preserve            | Advisory-by-design warnings; the parallel doctor check is drift-tested by §1.3's harness.                                                                       |
+
+### 1.8 Pending workstream additions (branch `feat/ri-050-publish-gate` @ `46784c8d`)
+
+| check                           | location                                             | kind | what it actually verifies                                                                                                                                                                                                                                                                          | enforcement point                                               | canonical check                     | disposition | rationale                                                                                      |
+| ------------------------------- | ---------------------------------------------------- | ---- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------- | ----------------------------------- | ----------- | ---------------------------------------------------------------------------------------------- |
+| Canonical terminal verification | `scripts/verify-release.mjs` (`pnpm verify:release`) | ts   | One command replaying the full mandatory set as stages — sanitization, upgrade-guard, typecheck (incl. preflight), lint, format, test, build — mirroring `ci.yml` step-for-step; fail-fast on first failing command; requires `bash`+`rsync` on PATH; `--stage <name>` for wiring smoke-tests only | `publish.yml#verify` (pending); `local` (`pnpm verify:release`) | QC-11 Terminal release verification | preserve    | The RI-N1 canonical command — CI and publication share one semantic checklist by construction. |
+| Verify-parity contract test     | `scripts/verify-release.test.mjs`                    | ts   | Parses the real `ci.yml`/`publish.yml`: stage table must match ci.yml step-for-step; every publish-effect step (name `publish*` or image-pushing) must transitively depend on `verify`; commit-identity assertion must be present; `verify` must carry no path filter                              | `test:checkout` → `CI ci.yml#test` (once merged)                | QC-11 Terminal release verification | preserve    | Guard-of-the-guard at checkout time — the two surfaces cannot drift apart silently.            |
+
+## 2. Canonical check set
+
+The deduplicated checks every row above maps onto. IDs are stable for RI-3-002 to consume.
+
+- **QC-1 Checkout integrity.** Owns: the checkout can run its gates — frozen-lockfile dependency resolution, required gate binaries present, no stale build lock, and the `apps/web/.next` generated-state trust chain (real directory, uid ownership, certified source fingerprint, certified symlink manifest). Implemented by `scripts/preflight.mjs` + frozen install steps.
+- **QC-2 Workspace typecheck.** Owns workspace-wide TypeScript soundness: per-package `tsc --noEmit` over built dependencies (`turbo typecheck`). The single definition invoked by CI, pre-push, and terminal verification.
+- **QC-3 Workspace lint.** Owns static-analysis policy: per-package ESLint under the root config. One config, one task, every surface.
+- **QC-4 Format check.** Owns formatting uniformity: Prettier check with the repo ignore list. (The pre-commit variant additionally fixes; the verdict form is this check.)
+- **QC-5 Test suite execution.** Owns execution of all test surfaces: checkout script units (`node --test`), per-package Vitest suites (including the framework shell chain and its python unitests), the installer-lane shim test, and — once wired — `test-roster-schema.py` and container E2E. Also owns guards-of-the-gate that live inside the chain (terminal-green contract, RCE regression).
+- **QC-6 Workspace build.** Owns artifact buildability: `turbo build` producing the artifacts publication consumes.
+- **QC-7 Framework sanitization.** Owns the open-source guarantee for the shipped framework package: no operator-identity tokens anywhere (examples included), no private `$HOME` defaults in shipped scripts, with a self-test that keeps the regexes honest.
+- **QC-8 Resident-context budget.** Owns the line-count ceilings on framework files injected into every agent's context (Constitution, dispatcher, RUNTIME.md slices) — the CI-enforceable half of the resident-prompt budget.
+- **QC-9 Test-membership enumeration.** Owns the property that no test suite can silently fall out of CI: disk population vs parsed enumeration surfaces, both-directions staleness, and signed exclusions with reasons. Includes its needle/control harness.
+- **QC-10 Upgrade/install safety.** Owns the #791 family: operator-path byte-identity across keep-mode upgrades (manifest guard), mid-failure rollback (errtrace-proven), durable pre-update snapshot + verify net + CWE-59 leaf guard, and the v2→v3 migration matrix with shell/TS parity.
+- **QC-11 Terminal release verification.** Owns the RI-N1 exact-commit binding: commit-identity assertion plus one canonical command (`pnpm verify:release`) replaying the complete mandatory set, with every publish effect depending on it; plus the checkout-time parity/DAG contract test that keeps pipeline and command in sync.
+- **QC-12 Publish-effect integrity.** Owns publication correctness: npm publish error classification (only already-published tolerated), next-lane versioning and post-publish resolution proof, and image destination/tag policy.
+- **QC-13 Staged-change hygiene.** Owns commit-time hygiene on staged files (prettier/eslint fix-and-restage) and the self-verifying hook wiring that guarantees the gates are actually installed.
+- **QC-14 Pre-push gate.** Owns the local push composition: preflight + typecheck + lint + format:check (tests deliberately deferred to CI).
+- **QC-15 Lease-gate architecture invariant.** Owns "no ungated runtime launches in production code": the scan + allowlist over `packages/`, `apps/`, `plugins/`, `tools/`.
+- **QC-16 Agent-runtime edit-time checks.** Owns edit-time feedback on agent hosts: the deps-preflight legibility sentinel and typecheck-on-edit, plus their regression harnesses.
+- **QC-17 Lease-enforcement wiring safety.** Owns the #869 C1/C2/C5 trio: activation capability probe (versioned contract), enforcement-hook wiring gate (default-deny with explicit opt-out), and the doctor check that surfaces a bricked host — with their shell/TS harnesses.
+- **QC-18 Operator-host drift audit.** Owns host-state health CI cannot see: `mosaic doctor` drift audit (+ fleet transport, both implementations), `fleet doctor` roster classification, `gateway doctor`/`gateway verify` service health, and installer advisories. Advisory exits are part of the contract.
+- **QC-19 Downstream rails presence check.** Owns "does a scaffolded project still carry its rails files" — today the TS `quality-rails check/doctor` presence loop; per RI-N4 this is the seed that must become the typed evaluator (presence alone is explicitly not parity).
+- **QC-20 Downstream enforcement verification.** Owns "do the rails actually block" on scaffolded projects: the behavioral planted-commit probe (type error, `any`, lint, gitleaks secret) currently in `verify.sh`/`verify.ps1` behind the `mosaic-quality-verify` adapter.
+- **QC-21 Downstream rails scaffolding.** Owns putting rails files into a target project: the shell template installer (+ PowerShell twin) and the TS `quality-rails init` scaffolder — currently two paths that must converge.
+
+## 3. Coverage gaps
+
+Enforced nowhere but implied, or named in docs/tooling but not wired:
+
+1. **Publication not yet bound to verification on `next`.** At this base (`8199261c`), `publish.yml` publish steps depend on `build` only; the `verify` step and `scripts/verify-release.mjs` exist on `feat/ri-050-publish-gate` (`46784c8d`) but are not merged. Until RI-1-001 lands, AC-RI-1's negative control cannot hold on the real pipeline.
+2. **Playwright E2E unwired.** `apps/web` ships `test:e2e` (`playwright test`) with real suites (`admin/auth/chat/navigation.spec.ts`); neither `pnpm test` nor any CI step invokes it. The web UI's user flows are verified only when an operator runs them manually.
+3. **No secret scanning on this repo.** The framework's own template pre-commit makes gitleaks **required**, and `verify.sh` proves detection with a planted key — but this repository's `.husky/pre-commit` (lint-staged only) and CI run no secret scan. The repo ships the control it does not use.
+4. **No dependency audit.** The quality `.woodpecker.yml` templates and `docs/CI-SETUP.md` specify `npm audit --audit-level=high` as a pipeline stage; nothing equivalent runs for this repo.
+5. **No coverage thresholds.** Templates enforce 80% Jest coverage thresholds; this repo's Vitest configs collect coverage with no thresholds — coverage is measured nowhere and enforced nowhere.
+6. **`test-roster-schema.py` invisible.** A real jsonschema regression suite wired to no surface and invisible to the enumeration guard (its population is `*.sh`; the suite is `.py`). Either enumerate it or sign an exclusion — silence here is the #1017 defect shape.
+7. **Presence-checker expectations ≠ this repo.** `quality-rails check` expects `.eslintrc`, `biome.json`, `.githooks/pre-commit`, `PR-CHECKLIST.md` for node projects — none describe this monorepo (husky, flat eslint config, no biome, no PR-CHECKLIST.md). The evaluator's check set must be per-subject (versioned, digested), not one global file list.
+8. **Chain-ordering residual (documented).** `test:framework-shell` is one `&&` chain: a failing link skips every later suite while the step still fails (measured in #1270 — four suites after position 44 had not run since a prior merge). The enumeration guard proves naming, not reachability; both residuals are in-file documented but structurally unfixed.
+9. **Signed-exclusion burndown open.** 16 signed exclusions remain in `test-enumeration-exclusions.txt`; several are "unmeasured in CI image" or blocked on missing CI tooling (tmux, setsid) — tracked under #1017/#1271. Each is an enforcement promise deferred, not delivered.
+10. **Windows twins unexercised.** `verify.ps1`, `install.ps1`, `mosaic-doctor.ps1` have no runner anywhere (no Windows CI); behavioral drift from their bash twins is undetectable by construction.
+11. **QA hook name vs behavior.** `qa-hook-handler.sh` files remediation report templates but performs no verification; the seam's actual gate value is only the deps-preflight sentinel. Anything relying on "QA automation hook" as a check is relying on report-filing.
+12. **Two test paths, one gated.** CI runs tests against ci-postgres (`DATABASE_URL` set); the local PGlite path is the documented default (AGENTS.md) until KBN-101-02/101-05. Only the CI path is enforced by pipeline.
+
+## 4. Disposition summary
+
+| disposition         | rows | checks                                                                                                                                                                                                                                                                                    |
+| ------------------- | ---- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| preserve            | 43   | Every canonical owner (QC-1..QC-18) plus correct guards-of-the-guard and thin adapters: all of §1.1, the CI-invoked framework probes and adapters in §1.2, all of §1.3, the C1/C2/C5 trio and doctors in §1.4, all of §1.5, all pipeline-only steps in §1.6, §1.7 rows 1 and 3, and §1.8. |
+| strengthen          | 2    | `quality-rails check` and `quality-rails doctor` (QC-19) — the RI-N4 evaluator seed: typed verdicts, versioned/digested check definitions, per-subject check sets.                                                                                                                        |
+| strengthen (review) | 9    | `verify.sh` + `verify.ps1` (QC-20), quality `install.sh`/`install.ps1` + `quality-rails init` (QC-21 — scaffold-path convergence), `test-roster-schema.py` (QC-5 — wire or sign), `qa-hook-stdin.sh` seam + `typecheck-hook.sh` (QC-16), `tools/e2e-install-test.sh` (QC-5 — CI lane).    |
+| retire              | 0    | None meet the bar: RI-N4 requires effective shell probes be **absorbed before** their paths retire, and no absorption exists yet. The `strengthen (review)` rows are the retirement candidates for RI-3-002 once the evaluator owns their behavior.                                       |
+
+Row total: 54. Canonical checks: 21 (QC-1..QC-21).

docs/reports/quality/1099-pipefail-sweep.md

@@ -6,78 +6,78 @@ This is a site inventory, not a risk count. `FIXED` means the early-exiting cons
 
 ## Tranche 1 — runtime and general scripts
 
-| Baseline site | Verdict | Construction / reason |
-| --- | --- | --- |
-| `tools/matrix-presence-harness/run.sh:38` | FIXED | nullglob array selects the first path; no pipeline |
-| `tools/e2e-install-test.sh:139` | FIXED | capture help completely, then grep via redirection |
-| `tools/install.sh:312` | FIXED | NUL `mapfile` reads all roots; count != 1 reaches the named malformed-archive diagnostic |
-| `scripts/analysis/reflect-board-history.sh:76` | FIXED | capture Git history completely, then grep via redirection |
-| `scripts/analysis/reflect-git-history.sh:67` | FIXED | grep reads from a here-string |
-| `scripts/analysis/reflect-git-history.sh:69` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/authentik/user-create.sh:72` | FIXED | jq `first(...)` reads the response directly |
-| `packages/mosaic/framework/tools/git/mutate-push-guard.sh:87` | FIXED | grep `-m1` reads the file directly; downstream `cut` consumes its complete scalar output |
-| `packages/mosaic/framework/tools/orchestrator/session-resume.sh:94` | FIXED | `mapfile` plus bounded indexed loop replaces `head` pipeline |
-| `packages/mosaic/framework/tools/prdy/prdy-status.sh:69` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:172` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:173` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:174` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:175` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:176` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:177` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:178` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/qa/typecheck-hook.sh:16` | FIXED | Bash regex extracts the first field without a pipeline |
-| `packages/mosaic/framework/tools/qa/typecheck-hook.sh:56` | FIXED | grep and bounded sed each read from a here-string |
-| `packages/mosaic/framework/tools/tmux/send-message.sh:113` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/tmux/send-message.sh:124` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/wake/detector.sh:126` | FIXED | one awk reads the manifest directly and exits after the first exact key |
-| `packages/mosaic/framework/tools/wake/detector.sh:270` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/wake/detector.sh:278` | FIXED | grep reads from a here-string |
-| `packages/mosaic/framework/tools/wake/digest.sh:647` | FIXED | capture complete locator output, then select first line by parameter expansion |
-| `packages/mosaic/framework/tools/wake/reconcile.sh:149` | FIXED | one awk reads the manifest directly and exits after the first exact key |
+| Baseline site                                                       | Verdict | Construction / reason                                                                    |
+| ------------------------------------------------------------------- | ------- | ---------------------------------------------------------------------------------------- |
+| `tools/matrix-presence-harness/run.sh:38`                           | FIXED   | nullglob array selects the first path; no pipeline                                       |
+| `tools/e2e-install-test.sh:139`                                     | FIXED   | capture help completely, then grep via redirection                                       |
+| `tools/install.sh:312`                                              | FIXED   | NUL `mapfile` reads all roots; count != 1 reaches the named malformed-archive diagnostic |
+| `scripts/analysis/reflect-board-history.sh:76`                      | FIXED   | capture Git history completely, then grep via redirection                                |
+| `scripts/analysis/reflect-git-history.sh:67`                        | FIXED   | grep reads from a here-string                                                            |
+| `scripts/analysis/reflect-git-history.sh:69`                        | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/authentik/user-create.sh:72`       | FIXED   | jq `first(...)` reads the response directly                                              |
+| `packages/mosaic/framework/tools/git/mutate-push-guard.sh:87`       | FIXED   | grep `-m1` reads the file directly; downstream `cut` consumes its complete scalar output |
+| `packages/mosaic/framework/tools/orchestrator/session-resume.sh:94` | FIXED   | `mapfile` plus bounded indexed loop replaces `head` pipeline                             |
+| `packages/mosaic/framework/tools/prdy/prdy-status.sh:69`            | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:172`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:173`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:174`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:175`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:176`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:177`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/reflect-stop-hook.sh:178`       | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/qa/typecheck-hook.sh:16`           | FIXED   | Bash regex extracts the first field without a pipeline                                   |
+| `packages/mosaic/framework/tools/qa/typecheck-hook.sh:56`           | FIXED   | grep and bounded sed each read from a here-string                                        |
+| `packages/mosaic/framework/tools/tmux/send-message.sh:113`          | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/tmux/send-message.sh:124`          | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/wake/detector.sh:126`              | FIXED   | one awk reads the manifest directly and exits after the first exact key                  |
+| `packages/mosaic/framework/tools/wake/detector.sh:270`              | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/wake/detector.sh:278`              | FIXED   | grep reads from a here-string                                                            |
+| `packages/mosaic/framework/tools/wake/digest.sh:647`                | FIXED   | capture complete locator output, then select first line by parameter expansion           |
+| `packages/mosaic/framework/tools/wake/reconcile.sh:149`             | FIXED   | one awk reads the manifest directly and exits after the first exact key                  |
 
 ## Explicit withdrawn / non-load-bearing sites
 
-| Baseline site | Verdict | Reason |
-| --- | --- | --- |
-| `tools/install.sh:182` | NOT-LOAD-BEARING | `|| true` explicitly discards lookup status |
-| `tools/install.sh:356` | UNREACHABLE-AND-WHY | `pnpm pack` writes one matching CLI tarball into a fresh directory immediately before lookup; citation withdrawn in #1099 |
-| `tools/install.sh:357` | UNREACHABLE-AND-WHY | same fresh-directory invariant for gateway tarball; citation withdrawn in #1099 |
-| `tools/install.sh:627` | NOT-LOAD-BEARING | `|| true` explicitly discards lookup status |
-| `scripts/agent/session-start.sh:70` | NOT-LOAD-BEARING | optional scratchpad lookup has `|| true` |
-| `packages/mosaic/framework/templates/repo/scripts/agent/session-start.sh:58` | NOT-LOAD-BEARING | optional scratchpad lookup has `|| true` |
-| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:25` | UNREACHABLE-AND-WHY | withdrawn in #1099 after designed-input reachability measurement; preserved without re-litigation |
-| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:27` | UNREACHABLE-AND-WHY | same withdrawn designed-input finding |
-| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:30` | UNREACHABLE-AND-WHY | same withdrawn designed-input finding |
-| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:32` | UNREACHABLE-AND-WHY | same withdrawn designed-input finding |
-| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:34` | UNREACHABLE-AND-WHY | same withdrawn designed-input finding |
+| Baseline site                                                                | Verdict             | Reason                                                                                                                    |
+| ---------------------------------------------------------------------------- | ------------------- | ------------------------------------------------------------------------------------------------------------------------- | --- | --------------------------------------- |
+| `tools/install.sh:182`                                                       | NOT-LOAD-BEARING    | `                                                                                                                         |     | true` explicitly discards lookup status |
+| `tools/install.sh:356`                                                       | UNREACHABLE-AND-WHY | `pnpm pack` writes one matching CLI tarball into a fresh directory immediately before lookup; citation withdrawn in #1099 |
+| `tools/install.sh:357`                                                       | UNREACHABLE-AND-WHY | same fresh-directory invariant for gateway tarball; citation withdrawn in #1099                                           |
+| `tools/install.sh:627`                                                       | NOT-LOAD-BEARING    | `                                                                                                                         |     | true` explicitly discards lookup status |
+| `scripts/agent/session-start.sh:70`                                          | NOT-LOAD-BEARING    | optional scratchpad lookup has `                                                                                          |     | true`                                   |
+| `packages/mosaic/framework/templates/repo/scripts/agent/session-start.sh:58` | NOT-LOAD-BEARING    | optional scratchpad lookup has `                                                                                          |     | true`                                   |
+| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:25`                     | UNREACHABLE-AND-WHY | withdrawn in #1099 after designed-input reachability measurement; preserved without re-litigation                         |
+| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:27`                     | UNREACHABLE-AND-WHY | same withdrawn designed-input finding                                                                                     |
+| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:30`                     | UNREACHABLE-AND-WHY | same withdrawn designed-input finding                                                                                     |
+| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:32`                     | UNREACHABLE-AND-WHY | same withdrawn designed-input finding                                                                                     |
+| `packages/mosaic/framework/tools/qa/qa-hook-stdin.sh:34`                     | UNREACHABLE-AND-WHY | same withdrawn designed-input finding                                                                                     |
 
 ## Tranche 2 — non-wake test harnesses
 
 All 22 baseline sites below are `FIXED`; the checked-in tranche fixture is passed through the same scanner and asserts all 22 occurrences and 21 normalized identities (the same response-split line occurs twice).
 
-| Baseline site(s) | Verdict | Construction |
-| --- | --- | --- |
-| `systemd/user/test-fleet-units.sh:148` | FIXED | capture tmux output, then grep via redirection |
-| `git/test-issue-comment-readback.sh:283,302` | FIXED | parameter expansion splits status/body without `head` |
-| `git/test-pr-review-gitea-comment.sh:228` | FIXED | parameter expansion splits status/body |
-| `git/test-lane-brief-pr-linkage.sh:72` | FIXED | grep reads from a here-string |
-| `git/test-pr-review-repo-host-override.sh:225-226` | FIXED | grep reads from a here-string |
-| `orchestrator/smoke-test.sh:67,72` | FIXED | parameter expansion selects first line |
-| `orchestrator/test-board-roll.sh:99-100` | FIXED | grep reads from a here-string |
-| `quality/scripts/test-upgrade-durable-snapshot.sh:180` | FIXED | complete sorted output is read with `mapfile`, then indexed |
-| `quality/scripts/test-upgrade-rollback.sh:339,356` | FIXED | direct `grep -m1` file reads; cleanup captures before testing |
-| `tmux/test-send-message-socket.sh:37,38,44-46,68,72` | FIXED | capture commands complete before redirected grep assertions |
-| `tmux/test-send-message-verdict.sh:34` | FIXED | grep reads from a here-string |
+| Baseline site(s)                                       | Verdict | Construction                                                  |
+| ------------------------------------------------------ | ------- | ------------------------------------------------------------- |
+| `systemd/user/test-fleet-units.sh:148`                 | FIXED   | capture tmux output, then grep via redirection                |
+| `git/test-issue-comment-readback.sh:283,302`           | FIXED   | parameter expansion splits status/body without `head`         |
+| `git/test-pr-review-gitea-comment.sh:228`              | FIXED   | parameter expansion splits status/body                        |
+| `git/test-lane-brief-pr-linkage.sh:72`                 | FIXED   | grep reads from a here-string                                 |
+| `git/test-pr-review-repo-host-override.sh:225-226`     | FIXED   | grep reads from a here-string                                 |
+| `orchestrator/smoke-test.sh:67,72`                     | FIXED   | parameter expansion selects first line                        |
+| `orchestrator/test-board-roll.sh:99-100`               | FIXED   | grep reads from a here-string                                 |
+| `quality/scripts/test-upgrade-durable-snapshot.sh:180` | FIXED   | complete sorted output is read with `mapfile`, then indexed   |
+| `quality/scripts/test-upgrade-rollback.sh:339,356`     | FIXED   | direct `grep -m1` file reads; cleanup captures before testing |
+| `tmux/test-send-message-socket.sh:37,38,44-46,68,72`   | FIXED   | capture commands complete before redirected grep assertions   |
+| `tmux/test-send-message-verdict.sh:34`                 | FIXED   | grep reads from a here-string                                 |
 
 ## Tranche 3 — wake validation harnesses
 
 All 26 baseline occurrences (25 normalized identities; one preimage selector occurs twice) are `FIXED` and mechanically bound through the wake fixture and shared scanner.
 
-| Baseline site(s) | Verdict | Construction |
-| --- | --- | --- |
-| `wake/test-wake-digest-quarantine.sh:567` | FIXED | complete match populations are captured, then first line selected by parameter expansion |
-| `wake/test-wake-preimage.sh:182-183,346-347` | FIXED | jq `first(...)` reads each JSONL file directly |
-| `wake/validate-973/microtest-wake-assert.sh:153,170-171,176,204-209,233-234,251-252,286-287` | FIXED | scalar assertions use here-strings; diagnostics use non-early sed ranges; source line captured before matching |
-| `wake/validate-973/validate-973.sh:110,119,180,182,187` | FIXED | scalar assertions use here-strings; diagnostic truncation uses consuming sed ranges |
+| Baseline site(s)                                                                             | Verdict | Construction                                                                                                   |
+| -------------------------------------------------------------------------------------------- | ------- | -------------------------------------------------------------------------------------------------------------- |
+| `wake/test-wake-digest-quarantine.sh:567`                                                    | FIXED   | complete match populations are captured, then first line selected by parameter expansion                       |
+| `wake/test-wake-preimage.sh:182-183,346-347`                                                 | FIXED   | jq `first(...)` reads each JSONL file directly                                                                 |
+| `wake/validate-973/microtest-wake-assert.sh:153,170-171,176,204-209,233-234,251-252,286-287` | FIXED   | scalar assertions use here-strings; diagnostics use non-early sed ranges; source line captured before matching |
+| `wake/validate-973/validate-973.sh:110,119,180,182,187`                                      | FIXED   | scalar assertions use here-strings; diagnostic truncation uses consuming sed ranges                            |
 
 The scoped inventory is complete: 26 runtime/general + 22 non-wake tests + 26 wake tests fixed; 11 explicitly withdrawn or non-load-bearing sites retain their documented verdicts.

docs/scratchpads/1179-required-security-di.md

@@ -0,0 +1,77 @@
+# #1179 — Required security DI wiring
+
+## Objective
+
+Eliminate the shared fail-open defect class **absence read as permission**:
+
+- FL-01: missing `CommandAuthorizationService` must refuse Nest startup and must not permit command effects.
+- FL-11: missing `SystemOverrideService` must refuse Nest startup and must not omit stored instruction authority while allowing provider/session effects.
+
+## Tracking
+
+- Issue: #1179, child of #1156
+- Branch: `fix/1179-required-security-di`
+- Base: `origin/next` at `216cd72226cd9ee17eea461cfe7cd0e010a22f02`
+
+## Plan
+
+1. RED: compile the real `AppModule` graph with each required provider independently removed, with a positive control for each intact binding.
+2. RED: directly exercise each malformed absence path and assert zero command/provider/session effects.
+3. Stop and report RED to the coordinator before production implementation.
+4. After authorization, make both constructor injections required, remove absence-as-permission branches, and update explicit legitimate optional test seams.
+5. Run focused Gateway tests, typecheck, lint, format, build, independent exact-head verification, and focused security review.
+
+## Immutable path fence
+
+Production changes are confined to:
+
+- `apps/gateway/src/commands/command-executor.service.ts`
+- `apps/gateway/src/agent/agent.service.ts`
+
+Tests and task evidence are confined to:
+
+- `apps/gateway/src/__tests__/required-security-wiring.test.ts`
+- existing direct-constructor specs that require explicit required arguments
+- `docs/scratchpads/1179-required-security-di.md`
+
+No files in #1178, #1072, #1080, or #1054 lanes are in scope. `docs/TASKS.md` is orchestrator-owned and will not be modified.
+
+## Budget
+
+No explicit token ceiling was provided. Working assumption: one narrow Gateway security packet; split and stop if either arm requires unrelated module rewiring.
+
+## Progress
+
+- Intake read from #1179 and parent #1156.
+- Base independently resolved from the issue's pre-native-stage ordering and repository `origin/next` ref; branch HEAD verified byte-for-byte against the remote ref.
+- Real consumers and direct constructors inventoried.
+
+## Tests
+
+### RED
+
+- `required-security-wiring.test.ts`: 4 failed, 2 passed before implementation.
+- Both real-graph negative controls showed module compilation accepted the missing target binding.
+- Direct FL-01 showed one unauthorized command effect; direct FL-11 showed one provider prompt and one session counter mutation.
+
+### GREEN
+
+- `required-security-wiring.test.ts`: 6/6 passed.
+- FL-01-only production revert: exactly the two FL-01 test cases failed; all four other cases, including FL-11, passed.
+- FL-11-only production revert: exactly the two FL-11 test cases failed; all four other cases, including FL-01, passed.
+- Full Gateway suite: 74 files passed, 7 skipped; 831 tests passed, 17 skipped.
+- Gateway typecheck: passed.
+- Gateway lint: passed.
+- Gateway build: passed.
+- Changed-file Prettier check: passed.
+
+### Review
+
+- Codex code review: APPROVE, 0 findings.
+- Codex focused security review: risk `none`, 0 findings.
+- Independent exact-head review remains assigned to Scrappy through the coordinator.
+
+## Risks / blockers
+
+- `AgentModule` / `CommandsModule` / `ChatModule` contain a production cycle; the module test therefore uses the real top-level `AppModule` and replaces only storage/network leaves, preserving the target service in each arm while isolating the separate required consumer that would otherwise mask that arm's defect.
+- No broad module rewrite was required.

docs/scratchpads/ri-050.md

@@ -0,0 +1,242 @@
+# Scratchpad — RI-050 orchestrator (jarvis, dragon-lin)
+
+Mission: alpha 0.0.50 release-integrity floor. Issue #1275. Base `next` @ 476db12b.
+Design SSOT: jarvis-brain `docs/plans/2026-08-16_mosaic-stack-sdlc-protocol.md` (SDLC-D-033..038).
+
+## Mode (Jason's directives)
+
+- Orchestrator: jarvis (this session, dragon-lin). NOT mos-claude; work stays on this host.
+- Workers: local pi headless — `pi --model zai/glm-5.3:high -p` in the card's worktree, tools read,bash,edit,write.
+- Delegation override of stack AGENTS.md `agent` column: rows carry `pi-glm-5.3` (outside cron table so no auto-claim).
+- Target branch: `next`. Cards branch from `origin/next`, squash-merge via PR.
+
+## Operational constraints (measured this session)
+
+- Main checkout at `/home/jwoltje/src/mosaic-stack` is a dirty diverged `main` (ahead 1139/behind 711) — NEVER touched. All work in `/home/jwoltje/src/mosaic-stack-worktrees/<branch>`.
+- Disk: /home 187G free. /tmp only 8.7G — keep pnpm stores/node_modules under /home.
+- `main` and `next` have DIVERGED; PRs target `next`.
+- Identity: pin `GITEA_LOGIN=mosaicstack-jarvis` for all wrapper ops. Issue #1275 verified authored by @jarvis.
+- `ci-queue-wait.sh` on this host is fail-open (board: fix #1032 not installed) — substitute SHA-status checks via `/commits/{sha}/status` and diff failing step names.
+- CI on PRs runs `pull_request` pipelines (any branch) incl. ci-postgres service. Push CI runs on main only; publish runs on push/tag to next + manual.
+- Wrapper gaps on this host per board (7 gaps; e.g. no pr-review-list, issue-assign broken, pr-merge makes no trailers): verify outcomes by reading back provider state, never trust rc alone.
+- Publish pipeline currently: install → build → publish-npm/publish-next-npm (+image). No verify. CI steps: install, sanitization, upgrade-guard, typecheck, lint, format, test, ci-postgres.
+
+## Budget
+
+Soft cap 250K. Projected 190K across 10 cards. Track per-card used vs estimate in TASKS.md notes.
+
+## Progress log
+
+- 2026-08-16 23:52 — Issue #1275 created (@jarvis verified).
+- 2026-08-16 23:5x — Bootstrap branch `docs/ri-050-mission-bootstrap` from origin/next@476db12b; PRD section + TASKS.md + this scratchpad written. RI-0-001 in-progress.
+
+## Wave 1 dispatched (2026-08-17 00:35)
+
+- RI-1-001 worker: pi glm-5.3:high, pid 2322125, worktree ri-1-001, log /var/tmp/ri-050/ri-1-001-run.log
+- RI-2-001 worker: pi glm-5.3:high, pid 2322126, worktree ri-2-001, log /var/tmp/ri-050/ri-2-001-run.log
+- Gotcha recorded: pi has no -f flag (that's pi-do.sh); pass brief as positional message. First launch died "Unknown option: -f" — relaunched.
+- CI lane: PR #1276 (bootstrap) fails `test` at base like every next PR — fred's green #1270 unblocks (comms sent 2026-08-17T05:21Z, `comms/20260817T052148Z__from-jarvis__650fe8.md`). Merge gate for all RI PRs queues behind #1270.
+- Live RI-N1 evidence posted to #1275 (comment 22915): pipeline 2439 publish-next-npm SUCCESS beside build-gateway FAILURE.
+
+---
+
+# HANDOFF — RI-050 continuation (written 2026-08-17 ~08:45 UTC, jarvis/dragon-lin)
+
+You are taking over the alpha 0.0.50 release-integrity workstream in place. Everything you
+need is on the remote. Read this whole file, then `docs/release-integrity/TASKS.md` (same
+branch), then the PRD section (`docs/PRD.md` § Release Integrity Workstream, same branch).
+
+## Identity / mode
+
+- Orchestrator identity: `jarvis` (dragon-lin). You continue as the RI-050 orchestrator under
+  whatever identity Jason gives you — if you are NOT jarvis, say so in comms and PR bodies.
+- Jason's standing directives for this mission: work happens on THIS repo (mosaicstack/stack),
+  PRs target `next` (NOT main), workers are local pi headless sessions on
+  `zai/glm-5.3:high`. Do not hand this to mos-claude. Do not borrow other seats' lanes.
+- All wrapper ops: pin `GITEA_LOGIN=mosaicstack-jarvis` (issue #1275 was verified authored by
+  @jarvis; keep identity consistent or verify yours with issue-view and READ BACK user.login).
+- CI substitution rule (this host's ci-queue-wait.sh is fail-open; fix #1032 not installed):
+  judge CI by SHA-status via `/api/v1/repos/mosaicstack/stack/commits/{sha}/status` or the
+  woodpecker API (`pipeline-status.sh -r mosaicstack/stack -n N -f json`), and DIFF THE
+  FAILING STEP NAMES rather than trusting rc.
+
+## Mission state at handoff
+
+Mission: alpha 0.0.50 release-integrity floor. Issue #1275 (open, has live-evidence comment).
+Decisions SDLC-D-033..038 live in jarvis-brain
+`docs/plans/2026-08-16_mosaic-stack-sdlc-protocol.md` (normative text also mirrored in the
+PRD section on this branch, so this repo is self-sufficient).
+
+Base: `origin/next` @ 476db12b. NOTE: `main` and `next` have DIVERGED — never base on main.
+
+Branches (all pushed, all clean trees):
+
+- `docs/ri-050-mission-bootstrap` @ 5114faa2 → PR #1276 (open, mergeable) — bootstrap docs +
+  this scratchpad + TASKS.md DAG. STATUS: CI red on `test` only, which is the known lane-wide
+  failure (see blocker below); own prettier issue already fixed.
+- `feat/ri-050-publish-gate` @ 0aa5ed35 → PR #1277 (open, mergeable) — RI-1-001 COMPLETE
+  (worker reported success, orchestrator review PASSED: verify step asserts CI_COMMIT_SHA ==
+  git rev-parse HEAD then runs canonical `pnpm verify:release`; every publish/image step
+  depends_on verify directly, confirmed by parsing the DAG: publish-npm, publish-next-npm,
+  build-gateway/appservice/web all -> [build, verify]; invariant test
+  scripts/verify-release.test.mjs passes 7/7 locally with negative fixtures). CI: same known
+  lane-red `test` step only.
+- `fix/ri-050-forge-fail-closed` @ 99b8f6ea → PR #1278 (open, mergeable) — RI-2-001 worker
+  reported success (typed `FORGE_*` capability errors, --simulate typed simulated everywhere,
+  vacuous true/echo gates replaced, closed ForgeOutcome set, 116 tests green incl. 16 new).
+  ORCHESTRATOR REVIEW NOT YET DONE — your first job. Review the diff
+  (1391 insertions across forge src), check the fail-closed paths and that simulated
+  results cannot satisfy any consumer, run `pnpm --filter @mosaicstack/forge test`.
+
+## The one blocker
+
+Every `next` PR pipeline is red on ONE assertion:
+`packages/mosaic/framework/tools/fleet/test-start-agent-session.sh:103` ("host provides 'pi'
+in the system path"). Pre-existing at base; affects PRs #1276/#1277/#1278 identically.
+fred's PR #1270 ("unblocks every PR on next") is green and open — it is HIS to merge; do not
+merge it yourself. jarvis sent comms (`comms/20260817T052148Z__from-jarvis__650fe8.md` in
+jarvis-brain) asking merge timing; no reply yet as of handoff. Merge gates for ALL RI PRs
+queue behind #1270 landing. Until then: review/develop freely, merge nothing that needs the
+green gate (docs-only #1276 arguably could merge red-lane with Jason's explicit call — ask,
+don't assume).
+
+## Remaining DAG (docs/release-integrity/TASKS.md is canonical)
+
+Wave 2 (next): RI-2-002 MACP fail-closed (brief pattern: mirror RI-2-001 for
+packages/macp/src/gate-runner.ts — empty commands, stub executors, unimplemented CI-provider
+gates fail closed; explicit simulate) and RI-4-001 PRD authority (one PRD service;
+@mosaicstack/prdy docs/prdy authoritative via `mosaic mission --plan`; `mosaic prdy` routes
+or becomes named Markdown adapter; mission<->PRD linkage persists — see PRD RI-N3).
+Wave 3: RI-3-001 probe inventory (docs), RI-5-001 web stale-safety.
+Wave 4: RI-1-002 negative-control tests, RI-3-002 TS evaluator absorbs shell probes.
+Final: RI-V-001 evidence pack (real green next publish run post-gate + all cards verified).
+
+## Worker mechanics (measured, reuse)
+
+- Dispatch: create worktree `git -C /home/jwoltje/src/mosaic-stack worktree add
+/home/jwoltje/src/mosaic-stack-worktrees/<id> -b <branch> origin/next`, write a brief to
+  /var/tmp/ri-050/, then run from INSIDE the worktree:
+  `pi -p --no-session --model zai/glm-5.3:high --tools read,bash,edit,write "$(cat brief.md)"`
+  (pi has NO -f flag — pass the brief as a positional message; first dispatch died on that).
+- Briefs for 1-001/2-001 are at /var/tmp/ri-050/ on dragon-lin (may not survive; the
+  pattern is fully described above and in TASKS.md).
+- Briefs must carry: worktree path, branch, base, requirements, known base-red list (so the
+  worker doesn't chase it), gates to run, PR creation command with GITEA_LOGIN pin, "do NOT
+  merge, do NOT touch docs/TASKS.md", and the JSON report format.
+- Verify worker claims: read the PR, run their tests yourself, parse pipeline step names.
+
+## Do-not-touch
+
+- Main checkout at /home/jwoltje/src/mosaic-stack (dirty diverged main) — never touch.
+- fred's open PRs (#1270 and others) — review evidence welcome, merging his is not yours.
+- Other RI PRs' authors' lanes: #1277/#1278 are yours to gate and merge ONCE lane is green
+  and review is recorded.
+- Never `--no-verify`; never bypass the wrapper-fails-closed rule (wrapper failure ⇒
+  `blocked + report exact command + stop`).
+
+## Session-restore command sequence
+
+1. `git -C /home/jwoltje/src/mosaic-stack-worktrees/ri-050 fetch origin --prune`
+2. Read this file + `docs/release-integrity/TASKS.md` + PRD section.
+3. Check PR states (#1270, #1276, #1277, #1278) and lane CI (SHA-status per above).
+4. Review RI-2-001 (PR #1278) if not yet done; then dispatch wave 2.
+
+— jarvis, 2026-08-17
+
+---
+
+# CONTINUATION — fargo (sb-it-1-dt)
+
+Orchestrator seat is now **fargo** on sb-it-1-dt (Jason, 2026-08-17): Claude seat, worktree discipline
+per fred's ruling (`~/agent-work/<slug>`, create → work → commit → push → remove as one act; the
+helper's `/src` refusal is a web1 convention, does not bind here). fred supports; lane rulings are
+his. Workers remain local pi `zai/glm-5.3:high` + limited Claude per Jason.
+
+## 2026-08-17 — RI-2-001 independent review DONE
+
+- **PR #1278 APPROVED** (Gitea review 172, pinned to head 99b8f6ea). Executed evidence, not read-only:
+  forge suite 116/116 at head (matches PR claim), forge lint green, forge typecheck green after
+  building `@mosaicstack/macp` dist (TS2307 on bare `pnpm install --frozen-lockfile` is a
+  minimal-install build-order artifact — the macp import is type-only, vitest passes unbuilt; CI
+  installs build workspace deps, hence green there), **workspace typecheck 45/45 at head**,
+  consumer sweep: no external type consumers of RunManifest/StageStatus/ForgeTaskResult/
+  TaskExecutor; only importer of the package is packages/mosaic via registerForgeCommand
+  (smoke test asserts registration/help only — cannot break). Digest gate (shaggy's) before==after
+  with both-arm reactivity controls.
+- CI red on #1276/#1277/#1278: lane-wide `test` failure only
+  (test-start-agent-session.sh:103, fred's guard mis-wired; #1270 unwires it). Fred measured log
+  content: one real byte-identical failure per pipeline (2456/2457/2458); 13 of ~14 `FAIL` grep
+  hits are passing fail-loud test NAMES. **The red carries no information about the RI changes.**
+- Non-blocking finding: README L141-143 + skills/mosaic-forge/SKILL.md document bare
+  `mosaic forge run`/`resume`, which now exits 1 FORGE_NO_EXECUTOR — fast-follow docs touch.
+- **Identity incident, ruled on by fred:** review 172 recorded under shared host principal
+  mos-dt-0, not fargo. Mechanism (measured, wrapper source): pr-review.sh resolves its acting login
+  from the tea login list only; no fargo tea login on this host → silent host-default fallback;
+  MOSAIC_GIT_IDENTITY is only read in detect-platform.sh get_gitea_token's fallback arm, never
+  reached. Exact-id read-back verifies against the writing token, so it passed while attribution
+  was wrong — durable-provenance machinery proves the write, not the seat. Fred's ruling: review
+  172 stands (substance/verdict/pin correct; label wrong); NO re-approval (one approval,
+  annotated, is the stronger record); fred posts the provenance correction under @fred with
+  --login fred-ms (hard-fail path); no fargo tea login ever (freeze + Jason's to authorize);
+  tooling gap filed by fred. Also explains (does not reopen) #1228's mos-dt-0 attribution.
+- Merge gate: all RI PRs queue behind fred's green #1270 (Jason's call).
+
+## Next
+
+1. Wave 2 dispatch: RI-2-002 (MACP fail-closed, mirror RI-2-001 pattern for
+   packages/macp/src/gate-runner.ts) + RI-4-001 (PRD authority). Two parallel workers max.
+2. Docs fast-follow (README + mosaic-forge skill) — fold into #1276 or a tiny docs card.
+3. RI-V-001 evidence at the end.
+
+— fargo, 2026-08-17
+
+---
+
+# RESUMPTION + DAILY-HANDOFF PROTOCOL (Jason, 2026-08-17)
+
+Orchestrator seat is back with **jarvis** (dragon-lin). Expect daily handoff between jarvis
+and fargo. Protocol (both seats, every handoff):
+
+1. **This file is the shared mission log.** Append a dated section per session: state
+   measured, actions taken, PR/review states, next actions. Never rewrite prior sections.
+2. **TASKS.md stays current within one session** — status, PR number in notes, review
+   evidence. Stale rows are handoff debt.
+3. **Cross-review rule (SDLC-D-011 in practice):** the reviewing seat must differ from the
+   producing seat. jarvis reviews fargo-dispatched PRs, fargo reviews jarvis-dispatched
+   PRs. Producers are always pi workers; dispatching seats verify before push; the other
+   seat records the Gitea review.
+4. Handoff = append here + push + (optional) issue #1275 comment if a decision changed.
+
+## RESUMED — jarvis/dragon-lin, 2026-08-17 (afternoon)
+
+- Measured: next = 8199261c (#1270 merged — lane unblocked for new PRs). #1293/#1294
+  (fargo, wave 2) CI-green, mergeable, no recorded reviews. #1276/#1277/#1278 still based
+  on 476db12b with stale red CI → need rebase onto 8199261c. #1278 review pinned to old
+  head 99b8f6ea by @mos-dt-0 (fargo's, mis-attributed per his note) — rebase will dismiss
+  it; re-approval must come from fargo/fred (author is @jarvis, cannot self-approve).
+- Live evidence #2: push pipeline 2462 (the #1270 merge itself) ran publish-next-npm
+  SUCCESS beside build-gateway FAILURE again.
+- Plan: rebase the three original branches; independently review #1293/#1294; merge order
+  once green+reviewed: #1276 (docs) → #1277 (publish gate) → #1278/#1293/#1294 (code).
+  After #1277 merges, watch the next push pipeline prove the verify gate live.
+- fargo's non-RI PRs (#1291/#1296/#1297/#1281) stay strictly his lane.
+
+## jarvis session 2026-08-17 (evening) — reviews, rebases, merge plan
+
+- Rebased #1276/#1277/#1278 onto 8199261c (heads 59e2c460 / 46784c8d / 4917df1f);
+  invariant tests 7/7 and forge 116/116 re-run green at new heads. #1270 touched
+  test-enumeration-exclusions.txt + package.json, NOT ci.yml — no semantic overlap with
+  #1277's ci.yml changes (checked, was a real concern).
+- Independent reviews recorded: #1293 APPROVED (review 173; macp 109/109; fail-closed paths
+  + aggregate state machine verified), #1294 APPROVED (review 174; prdy 20/20 + command
+  specs 9/9; single-writer + linkage persistence + labeled export + conflict-aware import
+  verified). Note: 19 unrelated mosaic suites fail on bare minimal install (known workspace
+  build-order artifact, documented by fargo) — not this change.
+- Measured: `next` has NO branch protection (API: only main listed). Cross-seat review
+  discipline is protocol-enforced, not Gitea-enforced. Flagged to fargo for Jason: direct
+  pushes to next trigger ungated publishes; protection is Jason's call (#1231 adjacent).
+- Merge order planned: #1276 (docs-only — no publish run) -> #1277 (first gated publish)
+  -> #1278 -> #1293 -> #1294. Sent fargo review requests with pinned head SHAs
+  (comms/20260818T011932Z__from-jarvis__a9c02b.md). Not merging #1293/#1294 before my three
+  clear fargo's review — order optimality beats speed; every pre-#1277 merge publishes ungated.
+- CI on the three rebased heads: pending at time of this entry.

eslint.config.mjs

@@ -34,6 +34,7 @@ export default tseslint.config(
             'packages/storage/vitest.config.ts',
             'packages/mosaic/vitest.config.ts',
             'packages/mosaic/__tests__/*.ts',
+            'packages/forge/__tests__/*.ts',
             'tools/federation-harness/*.ts',
           ],
         },

package.json

@@ -9,6 +9,7 @@
     "preflight": "node scripts/preflight.mjs",
     "clean:generated": "node scripts/clean-generated.mjs",
     "typecheck": "pnpm preflight && turbo run typecheck",
+    "verify:release": "node scripts/verify-release.mjs",
     "test:checkout": "node --test scripts/*.test.mjs",
     "test": "pnpm test:checkout && turbo run test && pnpm run test:installer",
     "test:installer": "bash tools/install-next-lane.test.sh",

packages/forge/PLAN.md

@@ -539,3 +539,43 @@ Not every brief needs full Board of Directors review. The classification system
 ### Backward compatibility
 
 Existing briefs without a `class` field are auto-classified. The default (no matching keywords) is `strategic`, so all existing runs get the full pipeline unless keywords trigger `technical`.
+
+---
+
+## Fail-Closed Execution & Explicit Simulation (SDLC-D-035)
+
+**Added:** 2026-08-17
+
+Forge fails closed when a required capability is missing. It never runs a
+pipeline with a stub executor and reports success.
+
+### Normal mode (default)
+
+- No task executor wired → the CLI exits nonzero with the typed capability
+  error `FORGE_NO_EXECUTOR`. No run is created.
+- A stage whose gate is approval-based (board approval, planning approvals,
+  remediation re-review, discovery/analysis attestations) records a typed
+  `waiting-for-authority` stage result and raises `FORGE_AUTHORITY_REQUIRED`.
+  It never passes vacuously.
+- A stage whose gate requires an unwired provider (AI reviewer, CI pipeline)
+  records a typed `blocked` stage result and raises `FORGE_NO_REVIEWER` /
+  `FORGE_NO_CI_PIPELINE`. The synthetic echo-review approval in `06-review`
+  and all vacuous `true` gates were removed.
+
+### Explicit simulation (`--simulate`)
+
+Opts into stub/synthetic execution. Every stage result, every gate result, and
+the run manifest carry the distinct typed status `simulated` (manifest also
+records `mode: "simulated"`). `simulated` is a non-satisfying outcome:
+`isSatisfyingOutcome()` and all completion/gate consumers treat only `passed`
+as satisfying. The CLI exits 0 for a simulated run only because the caller
+explicitly passed `--simulate`, and prints a loud SIMULATED banner.
+
+### Typed outcome model
+
+Every gate/task outcome is one of the closed set
+`passed | failed | blocked | error | waiting-for-authority | simulated |
+not-applicable`, with the reason recorded on the stage status and each gate
+result in `manifest.json`. Missing implementations, missing gate evidence,
+unknown stages, process errors, and timeouts map to fail-closed members —
+never to `passed`.

packages/forge/__tests__/fail-closed.test.ts

@@ -0,0 +1,319 @@
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
+import { describe, it, expect, beforeEach, afterEach } from 'vitest';
+
+import { generateBoardTasks } from '../src/board-tasks.js';
+import { STAGE_SPECS } from '../src/constants.js';
+import { ForgeCapabilityError } from '../src/errors.js';
+import {
+  evaluateStageGates,
+  gateLabel,
+  isCommandGate,
+  isSatisfyingOutcome,
+} from '../src/outcomes.js';
+import { loadManifest, runPipeline } from '../src/pipeline-runner.js';
+import type { ForgeTask, ForgeTaskResult, TaskExecutor } from '../src/types.js';
+
+/**
+ * Mock real executor that returns typed results.
+ *
+ * Command gates are "verified" by the mock so normal-mode runs can pass
+ * mechanically gated stages; authority/provider gates are never reported
+ * because they have no mechanical implementation.
+ */
+function createTypedExecutor(options?: {
+  failStage?: string;
+  gateOutcomes?: Record<string, 'passed' | 'failed' | 'simulated' | 'error' | 'blocked'>;
+}): TaskExecutor & { submittedTasks: ForgeTask[] } {
+  const submittedTasks: ForgeTask[] = [];
+  return {
+    submittedTasks,
+    async submitTask(task: ForgeTask) {
+      submittedTasks.push(task);
+    },
+    async waitForCompletion(taskId: string): Promise<ForgeTaskResult> {
+      const task = submittedTasks.find((t) => t.id === taskId);
+      const stageName = task?.metadata?.['stageName'] as string | undefined;
+
+      if (options?.failStage && stageName === options.failStage) {
+        return {
+          task_id: taskId,
+          outcome: 'failed',
+          reason: 'mock task failure',
+          completed_at: new Date().toISOString(),
+          exit_code: 1,
+          gate_results: [],
+        };
+      }
+
+      const gateResults = (task?.qualityGates ?? [])
+        .filter((gate) => isCommandGate(gate))
+        .map((gate) => {
+          const label = gateLabel(gate);
+          const outcome = options?.gateOutcomes?.[label] ?? 'passed';
+          return {
+            gate: label,
+            outcome,
+            reason: outcome === 'passed' ? 'mock verified' : `mock gate outcome: ${outcome}`,
+          };
+        });
+
+      return {
+        task_id: taskId,
+        outcome: 'passed',
+        reason: 'mock verified',
+        completed_at: new Date().toISOString(),
+        exit_code: 0,
+        gate_results: gateResults,
+      };
+    },
+    async getTaskStatus() {
+      return 'completed' as const;
+    },
+  };
+}
+
+describe('fail-closed: no executor wired', () => {
+  let tmpDir: string;
+  let briefPath: string;
+
+  beforeEach(() => {
+    tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'forge-failclosed-'));
+    briefPath = path.join(tmpDir, 'brief.md');
+    fs.writeFileSync(briefPath, '# Fix bug\n\nA bugfix for lint cleanup.');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  it('throws a typed FORGE_NO_EXECUTOR capability error without --simulate', async () => {
+    await expect(
+      runPipeline(briefPath, tmpDir, {
+        // no executor, no simulate — must fail closed, never run with a stub
+        stages: ['00-intake'],
+      }),
+    ).rejects.toMatchObject({
+      name: 'ForgeCapabilityError',
+      code: 'FORGE_NO_EXECUTOR',
+      capability: 'task-executor',
+    });
+  });
+
+  it('does not create a run directory when failing closed on a missing executor', async () => {
+    try {
+      await runPipeline(briefPath, tmpDir, { stages: ['00-intake'] });
+    } catch {
+      // expected
+    }
+    expect(fs.existsSync(path.join(tmpDir, '.forge', 'runs'))).toBe(false);
+  });
+
+  it('completes with every result typed simulated when simulate is set', async () => {
+    const result = await runPipeline(briefPath, tmpDir, {
+      simulate: true,
+      stages: ['00-intake', '00b-discovery', '02-planning-1', '06-review'],
+    });
+
+    expect(result.manifest.mode).toBe('simulated');
+    expect(result.manifest.status).toBe('simulated');
+
+    for (const stage of result.stages) {
+      const stageStatus = result.manifest.stages[stage];
+      expect(stageStatus?.status, `stage ${stage}`).toBe('simulated');
+      expect(stageStatus?.status, `stage ${stage}`).not.toBe('passed');
+      expect(stageStatus?.reason, `stage ${stage}`).toBeTruthy();
+      for (const gateResult of stageStatus?.gateResults ?? []) {
+        expect(gateResult.outcome, `gate ${gateResult.gate} of ${stage}`).toBe('simulated');
+        expect(gateResult.outcome, `gate ${gateResult.gate} of ${stage}`).not.toBe('passed');
+      }
+    }
+
+    // The persisted manifest agrees.
+    const persisted = loadManifest(result.runDir);
+    expect(persisted.mode).toBe('simulated');
+    expect(persisted.status).toBe('simulated');
+    expect(persisted.stages['02-planning-1']?.status).toBe('simulated');
+  });
+});
+
+describe('fail-closed: typed outcome model', () => {
+  it('only passed satisfies the gate/dependency predicate', () => {
+    expect(isSatisfyingOutcome('passed')).toBe(true);
+    expect(isSatisfyingOutcome('failed')).toBe(false);
+    expect(isSatisfyingOutcome('blocked')).toBe(false);
+    expect(isSatisfyingOutcome('error')).toBe(false);
+    expect(isSatisfyingOutcome('waiting-for-authority')).toBe(false);
+    expect(isSatisfyingOutcome('simulated')).toBe(false);
+    expect(isSatisfyingOutcome('not-applicable')).toBe(false);
+  });
+
+  it('a simulated gate result cannot satisfy the stage gate evaluation', () => {
+    const evaluation = evaluateStageGates('05-coding', STAGE_SPECS['05-coding']!.qualityGates, {
+      task_id: 'FORGE-x-05',
+      outcome: 'passed',
+      reason: 'executor claims success',
+      completed_at: new Date().toISOString(),
+      exit_code: 0,
+      gate_results: [{ gate: 'pnpm lint', outcome: 'simulated', reason: 'simulated gate' }],
+    });
+    expect(isSatisfyingOutcome(evaluation.outcome)).toBe(false);
+    expect(evaluation.outcome).toBe('error');
+  });
+
+  it('a simulated task outcome cannot satisfy evaluation in normal mode', () => {
+    const evaluation = evaluateStageGates('00-intake', [], {
+      task_id: 'FORGE-x-00',
+      outcome: 'simulated',
+      reason: 'executor reported simulated',
+      completed_at: new Date().toISOString(),
+      exit_code: 0,
+      gate_results: [],
+    });
+    expect(isSatisfyingOutcome(evaluation.outcome)).toBe(false);
+  });
+
+  it('a missing gate result blocks the stage instead of passing vacuously', () => {
+    const evaluation = evaluateStageGates('05-coding', STAGE_SPECS['05-coding']!.qualityGates, {
+      task_id: 'FORGE-x-05',
+      outcome: 'passed',
+      reason: 'executor claims success',
+      completed_at: new Date().toISOString(),
+      exit_code: 0,
+      gate_results: [],
+    });
+    expect(evaluation.outcome).toBe('blocked');
+  });
+});
+
+describe('fail-closed: authority and provider gates', () => {
+  let tmpDir: string;
+  let briefPath: string;
+
+  beforeEach(() => {
+    tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'forge-authority-'));
+    briefPath = path.join(tmpDir, 'brief.md');
+    fs.writeFileSync(briefPath, '# Fix bug\n\nA bugfix for lint cleanup.');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  it.each(['02-planning-1', '03-planning-2', '04-planning-3', '07-remediate'])(
+    'planning/remediation stage %s yields waiting-for-authority (not passed) in normal mode',
+    async (stage) => {
+      const executor = createTypedExecutor();
+      let runDir: string | undefined;
+
+      try {
+        await runPipeline(briefPath, tmpDir, {
+          executor,
+          stages: [stage as string],
+        });
+        expect.unreachable('runPipeline should have failed closed');
+      } catch (err) {
+        expect(err).toBeInstanceOf(ForgeCapabilityError);
+        expect((err as ForgeCapabilityError).code).toBe('FORGE_AUTHORITY_REQUIRED');
+        runDir = path.join(tmpDir, '.forge', 'runs');
+      }
+
+      const runIds = fs.readdirSync(runDir!);
+      expect(runIds).toHaveLength(1);
+      const manifest = loadManifest(path.join(runDir!, runIds[0]!));
+      expect(manifest.stages[stage]?.status).toBe('waiting-for-authority');
+      expect(manifest.stages[stage]?.status).not.toBe('passed');
+      expect(manifest.status).toBe('waiting-for-authority');
+    },
+  );
+
+  it('review stage fails closed with a typed FORGE_NO_REVIEWER error in normal mode', async () => {
+    const executor = createTypedExecutor();
+
+    try {
+      await runPipeline(briefPath, tmpDir, {
+        executor,
+        stages: ['06-review'],
+      });
+      expect.unreachable('runPipeline should have failed closed');
+    } catch (err) {
+      expect(err).toBeInstanceOf(ForgeCapabilityError);
+      expect((err as ForgeCapabilityError).code).toBe('FORGE_NO_REVIEWER');
+      expect((err as ForgeCapabilityError).capability).toBe('reviewer');
+    }
+
+    const runsDir = path.join(tmpDir, '.forge', 'runs');
+    const runIds = fs.readdirSync(runsDir);
+    const manifest = loadManifest(path.join(runsDir, runIds[0]!));
+    expect(manifest.stages['06-review']?.status).toBe('blocked');
+    expect(manifest.stages['06-review']?.status).not.toBe('passed');
+    expect(manifest.status).toBe('failed');
+  });
+
+  it('review stage produces simulated results under --simulate', async () => {
+    const result = await runPipeline(briefPath, tmpDir, {
+      simulate: true,
+      stages: ['06-review'],
+    });
+
+    expect(result.manifest.mode).toBe('simulated');
+    expect(result.manifest.stages['06-review']?.status).toBe('simulated');
+    for (const gateResult of result.manifest.stages['06-review']?.gateResults ?? []) {
+      expect(gateResult.outcome).toBe('simulated');
+    }
+  });
+
+  it('deploy stage fails closed without a wired ci-pipeline provider in normal mode', async () => {
+    const executor = createTypedExecutor();
+
+    await expect(
+      runPipeline(briefPath, tmpDir, {
+        executor,
+        stages: ['09-deploy'],
+      }),
+    ).rejects.toMatchObject({
+      name: 'ForgeCapabilityError',
+      code: 'FORGE_NO_CI_PIPELINE',
+    });
+  });
+});
+
+describe('fail-closed: no vacuous gate commands remain', () => {
+  it('stage constants contain no echo/synthetic-approval, vacuous true, or empty gate commands', () => {
+    for (const [stageName, spec] of Object.entries(STAGE_SPECS)) {
+      for (const gate of spec.qualityGates) {
+        const serialized = JSON.stringify(gate);
+        // The echo-review synthetic approval must be gone.
+        expect(serialized, `stage ${stageName} gate ${serialized}`).not.toContain('echo');
+        expect(serialized, `stage ${stageName} gate ${serialized}`).not.toMatch(/"verdict"\s*:/);
+        expect(serialized, `stage ${stageName} gate ${serialized}`).not.toMatch(
+          /"summary"\s*:\s*"review-pass"/,
+        );
+        // No vacuous literal `true` gate.
+        expect(gate, `stage ${stageName}`).not.toBe('true');
+        // Command gates must carry a real, non-empty command.
+        if (isCommandGate(gate)) {
+          const command = typeof gate === 'string' ? gate : gate.command;
+          expect(command.trim().length, `stage ${stageName} gate ${serialized}`).toBeGreaterThan(0);
+        }
+      }
+    }
+  });
+
+  it('board tasks contain no vacuous true gates', () => {
+    const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'forge-board-gates-'));
+    try {
+      const tasks = generateBoardTasks('# Brief', [], tmpDir, 'BOARD-TEST');
+      for (const task of tasks) {
+        for (const gate of task.qualityGates) {
+          expect(gate, `task ${task.id}`).not.toBe('true');
+          const serialized = JSON.stringify(gate);
+          expect(serialized, `task ${task.id} gate ${serialized}`).not.toContain('echo');
+        }
+      }
+    } finally {
+      fs.rmSync(tmpDir, { recursive: true, force: true });
+    }
+  });
+});

packages/forge/__tests__/pipeline-runner.test.ts

@@ -12,10 +12,10 @@ import {
   resumePipeline,
   getPipelineStatus,
 } from '../src/pipeline-runner.js';
-import type { ForgeTask, RunManifest, TaskExecutor } from '../src/types.js';
-import type { TaskResult } from '@mosaicstack/macp';
+import type { ForgeTask, ForgeTaskResult, RunManifest, TaskExecutor } from '../src/types.js';
+import { gateLabel, isCommandGate } from '../src/outcomes.js';
 
-/** Mock TaskExecutor that records submitted tasks and returns success. */
+/** Mock TaskExecutor that records submitted tasks and returns typed results. */
 function createMockExecutor(options?: {
   failStage?: string;
 }): TaskExecutor & { submittedTasks: ForgeTask[] } {
@@ -25,7 +25,7 @@ function createMockExecutor(options?: {
     async submitTask(task: ForgeTask) {
       submittedTasks.push(task);
     },
-    async waitForCompletion(taskId: string): Promise<TaskResult> {
+    async waitForCompletion(taskId: string): Promise<ForgeTaskResult> {
       const failStage = options?.failStage;
       const task = submittedTasks.find((t) => t.id === taskId);
       const stageName = task?.metadata?.['stageName'] as string | undefined;
@@ -33,7 +33,8 @@ function createMockExecutor(options?: {
       if (failStage && stageName === failStage) {
         return {
           task_id: taskId,
-          status: 'failed',
+          outcome: 'failed',
+          reason: 'mock task failure',
           completed_at: new Date().toISOString(),
           exit_code: 1,
           gate_results: [],
@@ -41,10 +42,17 @@ function createMockExecutor(options?: {
       }
       return {
         task_id: taskId,
-        status: 'completed',
+        outcome: 'passed',
+        reason: 'mock verified',
         completed_at: new Date().toISOString(),
         exit_code: 0,
-        gate_results: [],
+        gate_results: (task?.qualityGates ?? [])
+          .filter((gate) => isCommandGate(gate))
+          .map((gate) => ({
+            gate: gateLabel(gate),
+            outcome: 'passed' as const,
+            reason: 'mock verified',
+          })),
       };
     },
     async getTaskStatus() {
@@ -156,12 +164,13 @@ describe('runPipeline', () => {
     const executor = createMockExecutor();
     const result = await runPipeline(briefPath, tmpDir, {
       executor,
-      stages: ['00-intake', '00b-discovery'],
+      stages: ['00-intake', '05-coding'],
     });
 
     expect(result.runId).toMatch(/^\d{8}-\d{6}$/);
-    expect(result.stages).toEqual(['00-intake', '00b-discovery']);
+    expect(result.stages).toEqual(['00-intake', '05-coding']);
     expect(result.manifest.status).toBe('completed');
+    expect(result.manifest.mode).toBe('normal');
     expect(executor.submittedTasks).toHaveLength(2);
   });
 
@@ -180,12 +189,17 @@ describe('runPipeline', () => {
     const executor = createMockExecutor();
     const result = await runPipeline(briefPath, tmpDir, {
       executor,
-      stages: ['00-intake', '00b-discovery'],
+      stages: ['00-intake', '05-coding'],
     });
 
     const manifest = loadManifest(result.runDir);
     expect(manifest.stages['00-intake']?.status).toBe('passed');
-    expect(manifest.stages['00b-discovery']?.status).toBe('passed');
+    expect(manifest.stages['05-coding']?.status).toBe('passed');
+    expect(manifest.stages['05-coding']?.gateResults?.map((g) => g.outcome)).toEqual([
+      'passed',
+      'passed',
+      'passed',
+    ]);
   });
 
   it('respects CLI class override', async () => {
@@ -215,7 +229,7 @@ describe('runPipeline', () => {
     const executor = createMockExecutor();
     await runPipeline(briefPath, tmpDir, {
       executor,
-      stages: ['00-intake', '00b-discovery', '02-planning-1'],
+      stages: ['00-intake', '05-coding', '08-test'],
     });
 
     expect(executor.submittedTasks[0]!.dependsOn).toBeUndefined();
@@ -224,14 +238,14 @@ describe('runPipeline', () => {
   });
 
   it('handles stage failure', async () => {
-    const executor = createMockExecutor({ failStage: '00b-discovery' });
+    const executor = createMockExecutor({ failStage: '05-coding' });
 
     await expect(
       runPipeline(briefPath, tmpDir, {
         executor,
-        stages: ['00-intake', '00b-discovery'],
+        stages: ['00-intake', '05-coding'],
       }),
-    ).rejects.toThrow('Stage 00b-discovery failed');
+    ).rejects.toThrow('Stage 05-coding failed');
   });
 
   it('marks manifest as failed on stage failure', async () => {
@@ -270,30 +284,143 @@ describe('resumePipeline', () => {
     fs.rmSync(tmpDir, { recursive: true, force: true });
   });
 
-  it('resumes from first incomplete stage', async () => {
-    // First run fails on discovery
-    const executor1 = createMockExecutor({ failStage: '00b-discovery' });
-    let runDir: string;
+  it('resumes from first incomplete stage and fails closed at the next provider gate', async () => {
+    // Simulate a run whose authority stages were approved out-of-band
+    // (recorded as passed) and whose coding stage failed mechanically.
+    const runId = '20260101-000000';
+    const runDir = path.join(tmpDir, '.forge', 'runs', runId);
+    fs.mkdirSync(runDir, { recursive: true });
+    const passed = { status: 'passed' as const, startedAt: '2026-01-01T00:00:00Z' };
+    saveManifest(runDir, {
+      runId,
+      brief: briefPath,
+      codebase: tmpDir,
+      briefClass: 'hotfix',
+      classSource: 'frontmatter',
+      forceBoard: false,
+      mode: 'normal',
+      createdAt: '2026-01-01T00:00:00Z',
+      updatedAt: '2026-01-01T00:00:00Z',
+      currentStage: '05-coding',
+      status: 'failed',
+      stages: {
+        '00-intake': passed,
+        '00b-discovery': passed,
+        '02-planning-1': passed,
+        '03-planning-2': passed,
+        '04-planning-3': passed,
+        '05-coding': { status: 'failed', reason: 'gate failed' },
+      },
+    });
 
-    try {
-      await runPipeline(briefPath, tmpDir, {
-        executor: executor1,
-        stages: ['00-intake', '00b-discovery', '02-planning-1'],
-      });
-    } catch {
-      // expected
+    // Resume re-runs 05-coding (the first non-passed stage), then fails
+    // closed at 06-review because no reviewer provider is wired.
+    const executor = createMockExecutor();
+    await expect(resumePipeline(runDir, executor)).rejects.toMatchObject({
+      name: 'ForgeCapabilityError',
+      code: 'FORGE_NO_REVIEWER',
+    });
+
+    const manifest = loadManifest(runDir);
+    expect(manifest.stages['05-coding']?.status).toBe('passed');
+    expect(manifest.stages['06-review']?.status).toBe('blocked');
+    expect(manifest.status).toBe('failed');
+  });
+
+  it('resumes to completion as simulated under explicit simulate', async () => {
+    const runId = '20260101-000003';
+    const runDir = path.join(tmpDir, '.forge', 'runs', runId);
+    fs.mkdirSync(runDir, { recursive: true });
+    const passed = { status: 'passed' as const, startedAt: '2026-01-01T00:00:00Z' };
+    saveManifest(runDir, {
+      runId,
+      brief: briefPath,
+      codebase: tmpDir,
+      briefClass: 'hotfix',
+      classSource: 'frontmatter',
+      forceBoard: false,
+      mode: 'normal',
+      createdAt: '2026-01-01T00:00:00Z',
+      updatedAt: '2026-01-01T00:00:00Z',
+      currentStage: '05-coding',
+      status: 'failed',
+      stages: {
+        '00-intake': passed,
+        '00b-discovery': passed,
+        '02-planning-1': passed,
+        '03-planning-2': passed,
+        '04-planning-3': passed,
+        '05-coding': { status: 'failed', reason: 'gate failed' },
+      },
+    });
+
+    const result = await resumePipeline(runDir, undefined, { simulate: true });
+
+    expect(result.manifest.status).toBe('simulated');
+    expect(result.manifest.mode).toBe('simulated');
+    expect(result.stages[0]).toBe('05-coding');
+    for (const stage of result.stages) {
+      expect(result.manifest.stages[stage]?.status).toBe('simulated');
     }
+  });
 
-    const runsDir = path.join(tmpDir, '.forge', 'runs');
-    runDir = path.join(runsDir, fs.readdirSync(runsDir)[0]!);
+  it('fails closed on resume when the next stage needs authority sign-off', async () => {
+    const runId = '20260101-000001';
+    const runDir = path.join(tmpDir, '.forge', 'runs', runId);
+    fs.mkdirSync(runDir, { recursive: true });
+    saveManifest(runDir, {
+      runId,
+      brief: briefPath,
+      codebase: tmpDir,
+      briefClass: 'hotfix',
+      classSource: 'frontmatter',
+      forceBoard: false,
+      mode: 'normal',
+      createdAt: '2026-01-01T00:00:00Z',
+      updatedAt: '2026-01-01T00:00:00Z',
+      currentStage: '00-intake',
+      status: 'in_progress',
+      stages: {
+        '00-intake': { status: 'passed' },
+      },
+    });
 
-    // Resume should pick up from 00b-discovery
-    const executor2 = createMockExecutor();
-    const result = await resumePipeline(runDir, executor2);
+    const executor = createMockExecutor();
+    await expect(resumePipeline(runDir, executor)).rejects.toMatchObject({
+      name: 'ForgeCapabilityError',
+      code: 'FORGE_AUTHORITY_REQUIRED',
+    });
 
-    expect(result.manifest.status).toBe('completed');
-    // Should have re-run from 00b-discovery onward
-    expect(result.stages[0]).toBe('00b-discovery');
+    const manifest = loadManifest(runDir);
+    expect(manifest.stages['00b-discovery']?.status).toBe('waiting-for-authority');
+    expect(manifest.status).toBe('waiting-for-authority');
+  });
+
+  it('fails closed on resume without an executor or --simulate', async () => {
+    const runId = '20260101-000002';
+    const runDir = path.join(tmpDir, '.forge', 'runs', runId);
+    fs.mkdirSync(runDir, { recursive: true });
+    saveManifest(runDir, {
+      runId,
+      brief: briefPath,
+      codebase: tmpDir,
+      briefClass: 'hotfix',
+      classSource: 'frontmatter',
+      forceBoard: false,
+      mode: 'normal',
+      createdAt: '2026-01-01T00:00:00Z',
+      updatedAt: '2026-01-01T00:00:00Z',
+      currentStage: '00-intake',
+      status: 'in_progress',
+      stages: {
+        '00-intake': { status: 'passed' },
+      },
+    });
+
+    await expect(resumePipeline(runDir)).rejects.toMatchObject({
+      name: 'ForgeCapabilityError',
+      code: 'FORGE_NO_EXECUTOR',
+    });
   });
 });
 

packages/forge/src/board-tasks.ts

@@ -95,7 +95,14 @@ export function generateBoardTasks(
       briefPath,
       resultPath: resultRelPath,
       timeoutSeconds: 120,
-      qualityGates: ['true'],
+      qualityGates: [
+        {
+          kind: 'authority',
+          capability: 'board-approval',
+          reason:
+            'persona evaluation is judged by board synthesis (authority review); no mechanical gate exists',
+        },
+      ],
       metadata: {
         personaName: persona.name,
         personaSlug: persona.slug,
@@ -121,7 +128,13 @@ export function generateBoardTasks(
     timeoutSeconds: 120,
     dependsOn: personaTaskIds,
     dependsOnPolicy: 'all_terminal',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'board-approval',
+        reason: 'board synthesis is an authority decision; no mechanical gate exists',
+      },
+    ],
     metadata: {
       resultOutputPath: synthesisResult,
       inputResultPaths: personaResultPaths,

packages/forge/src/cli.spec.ts

@@ -1,7 +1,11 @@
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
 import { Command } from 'commander';
-import { describe, expect, it } from 'vitest';
+import { describe, expect, it, vi, beforeEach, afterEach } from 'vitest';
 
 import { registerForgeCommand } from './cli.js';
+import { loadManifest } from './pipeline-runner.js';
 
 describe('registerForgeCommand', () => {
   it('registers a "forge" command on the parent program', () => {
@@ -55,3 +59,94 @@ describe('registerForgeCommand', () => {
     }).not.toThrow();
   });
 });
+
+describe('forge run fail-closed behavior (SDLC-D-035)', () => {
+  let tmpDir: string;
+  let briefPath: string;
+  let errSpy: ReturnType<typeof vi.spyOn>;
+  let logSpy: ReturnType<typeof vi.spyOn>;
+  let prevExitCode: string | number | null | undefined;
+
+  const parse = (args: string[]) => {
+    const program = new Command();
+    registerForgeCommand(program);
+    return program.parseAsync(['forge', ...args], { from: 'user' });
+  };
+
+  beforeEach(() => {
+    tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'forge-cli-failclosed-'));
+    briefPath = path.join(tmpDir, 'brief.md');
+    fs.writeFileSync(briefPath, '# Fix bug\n\nA bugfix for lint cleanup.');
+    errSpy = vi.spyOn(console, 'error').mockImplementation(() => {});
+    logSpy = vi.spyOn(console, 'log').mockImplementation(() => {});
+    prevExitCode = process.exitCode;
+  });
+
+  afterEach(() => {
+    errSpy.mockRestore();
+    logSpy.mockRestore();
+    process.exitCode = prevExitCode;
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  it('exits nonzero with a typed FORGE_NO_EXECUTOR error when no executor is wired and --simulate is absent', async () => {
+    await parse(['run', '--brief', briefPath, '--codebase', tmpDir]);
+
+    expect(process.exitCode).toBe(1);
+    const errText = errSpy.mock.calls.map((c) => c.join(' ')).join('\n');
+    expect(errText).toContain('FORGE_NO_EXECUTOR');
+    // It must never run the pipeline with a stub and report success.
+    expect(fs.existsSync(path.join(tmpDir, '.forge', 'runs'))).toBe(false);
+  });
+
+  it('completes with typed simulated results and exit 0 under explicit --simulate', async () => {
+    await parse(['run', '--brief', briefPath, '--codebase', tmpDir, '--simulate']);
+
+    expect(process.exitCode).toBeUndefined();
+
+    // Loud simulated-mode summary.
+    const logText = logSpy.mock.calls.map((c) => c.join(' ')).join('\n');
+    expect(logText).toContain('SIMULATED');
+
+    // Manifest records the mode and simulated per-result statuses.
+    const runsDir = path.join(tmpDir, '.forge', 'runs');
+    const runIds = fs.readdirSync(runsDir);
+    expect(runIds).toHaveLength(1);
+    const manifest = loadManifest(path.join(runsDir, runIds[0]!));
+    expect(manifest.mode).toBe('simulated');
+    expect(manifest.status).toBe('simulated');
+    for (const stageStatus of Object.values(manifest.stages)) {
+      expect(stageStatus?.status).toBe('simulated');
+      for (const gateResult of stageStatus?.gateResults ?? []) {
+        expect(gateResult.outcome).toBe('simulated');
+      }
+    }
+  });
+
+  it('resume exits nonzero with a typed FORGE_NO_EXECUTOR error without --simulate', async () => {
+    const runDir = path.join(tmpDir, '.forge', 'runs', '20260101-000000');
+    fs.mkdirSync(runDir, { recursive: true });
+    fs.writeFileSync(
+      path.join(runDir, 'manifest.json'),
+      JSON.stringify({
+        runId: '20260101-000000',
+        brief: briefPath,
+        codebase: tmpDir,
+        briefClass: 'hotfix',
+        classSource: 'frontmatter',
+        forceBoard: false,
+        createdAt: '2026-01-01T00:00:00Z',
+        updatedAt: '2026-01-01T00:00:00Z',
+        currentStage: '00-intake',
+        status: 'in_progress',
+        stages: { '00-intake': { status: 'passed' } },
+      }),
+    );
+
+    await parse(['resume', '20260101-000000', '--project', tmpDir]);
+
+    expect(process.exitCode).toBe(1);
+    const errText = errSpy.mock.calls.map((c) => c.join(' ')).join('\n');
+    expect(errText).toContain('FORGE_NO_EXECUTOR');
+  });
+});

packages/forge/src/cli.ts

@@ -5,37 +5,47 @@ import type { Command } from 'commander';
 
 import { classifyBrief } from './brief-classifier.js';
 import { STAGE_LABELS, STAGE_SEQUENCE } from './constants.js';
+import { ForgeCapabilityError } from './errors.js';
 import { getEffectivePersonas, loadBoardPersonas } from './persona-loader.js';
 import { generateRunId, getPipelineStatus, loadManifest, runPipeline } from './pipeline-runner.js';
-import type { PipelineOptions, RunManifest, TaskExecutor } from './types.js';
-
-// ---------------------------------------------------------------------------
-// Stub executor — used when no real executor is wired at CLI invocation time.
-// ---------------------------------------------------------------------------
-
-const stubExecutor: TaskExecutor = {
-  async submitTask(task) {
-    console.log(`  [forge] stage submitted: ${task.id} (${task.title})`);
-  },
-  async waitForCompletion(taskId, _timeoutMs) {
-    console.log(`  [forge] stage complete:  ${taskId}`);
-    return {
-      task_id: taskId,
-      status: 'completed' as const,
-      completed_at: new Date().toISOString(),
-      exit_code: 0,
-      gate_results: [],
-    };
-  },
-  async getTaskStatus(_taskId) {
-    return 'completed' as const;
-  },
-};
+import { createSimulatedExecutor } from './simulated-executor.js';
+import type { PipelineOptions, RunManifest, RunMode } from './types.js';
 
 // ---------------------------------------------------------------------------
 // Helpers
 // ---------------------------------------------------------------------------
 
+/** Resolve a run's effective mode, defaulting legacy manifests to normal. */
+function runModeOf(manifest: RunManifest): RunMode {
+  return manifest.mode ?? 'normal';
+}
+
+/** Print a loud banner so a simulated run can never be misread as verified. */
+function printSimulatedBanner(): void {
+  console.log('');
+  console.log('[forge] ===============================================================');
+  console.log('[forge] MODE: SIMULATED — no stage or gate was really executed.');
+  console.log('[forge] All results are synthetic and MUST NOT be read as verified');
+  console.log('[forge] success. Wire a real executor/providers and re-run to verify.');
+  console.log('[forge] ===============================================================');
+}
+
+/** Print a typed error line for fail-closed capability errors. */
+function printCapabilityError(err: ForgeCapabilityError): void {
+  console.error(`[forge] error ${err.code}: ${err.message}`);
+  console.error(`[forge] missing capability: ${err.capability}`);
+}
+
+/** Handle a pipeline error uniformly: typed capability errors get their code. */
+function handlePipelineError(err: unknown): void {
+  if (err instanceof ForgeCapabilityError) {
+    printCapabilityError(err);
+  } else {
+    console.error(`[forge] pipeline failed: ${err instanceof Error ? err.message : String(err)}`);
+  }
+  process.exitCode = 1;
+}
+
 function formatDuration(startedAt?: string, completedAt?: string): string {
   if (!startedAt || !completedAt) return '-';
   const ms = new Date(completedAt).getTime() - new Date(startedAt).getTime();
@@ -44,19 +54,24 @@ function formatDuration(startedAt?: string, completedAt?: string): string {
 }
 
 function printManifestTable(manifest: RunManifest): void {
+  const mode = runModeOf(manifest);
   console.log(`\nRun ID : ${manifest.runId}`);
   console.log(`Status : ${manifest.status}`);
+  console.log(`Mode   : ${mode}`);
+  if (mode === 'simulated') {
+    console.log('WARNING: SIMULATED RUN — results are synthetic, not verified success.');
+  }
   console.log(`Brief  : ${manifest.brief}`);
   console.log(`Class  : ${manifest.briefClass} (${manifest.classSource})`);
   console.log(`Updated: ${manifest.updatedAt}`);
   console.log('');
-  console.log('Stage'.padEnd(22) + 'Status'.padEnd(14) + 'Duration');
-  console.log('-'.repeat(50));
+  console.log('Stage'.padEnd(22) + 'Status'.padEnd(24) + 'Duration');
+  console.log('-'.repeat(60));
   for (const stage of STAGE_SEQUENCE) {
     const s = manifest.stages[stage];
     if (!s) continue;
     const label = (STAGE_LABELS[stage] ?? stage).padEnd(22);
-    const status = s.status.padEnd(14);
+    const status = s.status.padEnd(24);
     const dur = formatDuration(s.startedAt, s.completedAt);
     console.log(`${label}${status}${dur}`);
   }
@@ -90,23 +105,58 @@ function listRecentRuns(projectRoot?: string): void {
   }
 
   console.log('\nRecent runs:');
-  console.log('Run ID'.padEnd(22) + 'Status'.padEnd(14) + 'Brief');
-  console.log('-'.repeat(70));
+  console.log('Run ID'.padEnd(22) + 'Status'.padEnd(24) + 'Mode'.padEnd(12) + 'Brief');
+  console.log('-'.repeat(80));
 
   for (const runId of entries) {
     const runDir = path.join(runsDir, runId);
     try {
       const manifest = loadManifest(runDir);
-      const status = manifest.status.padEnd(14);
+      const status = manifest.status.padEnd(24);
+      const mode = runModeOf(manifest).padEnd(12);
       const brief = path.basename(manifest.brief);
-      console.log(`${runId.padEnd(22)}${status}${brief}`);
+      console.log(`${runId.padEnd(22)}${status}${mode}${brief}`);
     } catch {
-      console.log(`${runId.padEnd(22)}${'(unreadable)'.padEnd(14)}`);
+      console.log(`${runId.padEnd(22)}${'(unreadable)'.padEnd(24)}`);
     }
   }
   console.log('');
 }
 
+/**
+ * Apply the exit-code policy for a finished pipeline run (SDLC-D-035):
+ *
+ * - exit 0 only for a verified `completed` normal run, or for an overall
+ *   `simulated` run when the caller explicitly passed --simulate;
+ * - anything else exits nonzero so it can never be read as success.
+ */
+function applyRunExitPolicy(result: { manifest: RunManifest; runDir: string }, simulate: boolean) {
+  const { manifest } = result;
+
+  if (runModeOf(manifest) === 'simulated') {
+    if (!simulate || manifest.status !== 'simulated') {
+      console.error(
+        '[forge] error FORGE_MODE_MISMATCH: run reports simulated results without an explicit, ' +
+          'consistent --simulate request; refusing to report success.',
+      );
+      process.exitCode = 1;
+      return;
+    }
+    printSimulatedBanner();
+    console.log(`[forge] run directory: ${result.runDir}`);
+    return; // exit 0 — the caller explicitly opted into simulation
+  }
+
+  if (manifest.status !== 'completed') {
+    console.error(`[forge] run did not complete: terminal status '${manifest.status}'`);
+    process.exitCode = 1;
+    return;
+  }
+
+  console.log(`[forge] pipeline complete (mode: normal): ${manifest.runId}`);
+  console.log(`[forge] run directory: ${result.runDir}`);
+}
+
 // ---------------------------------------------------------------------------
 // Register function
 // ---------------------------------------------------------------------------
@@ -129,6 +179,11 @@ export function registerForgeCommand(parent: Command): void {
     .option('--config <path>', 'Path to forge config file (.forge/config.yaml)')
     .option('--codebase <path>', 'Codebase root to pass to the pipeline', process.cwd())
     .option('--dry-run', 'Print planned stages without executing', false)
+    .option(
+      '--simulate',
+      'Simulate execution without real providers (every result is typed simulated, never verified)',
+      false,
+    )
     .action(
       async (opts: {
         brief: string;
@@ -137,6 +192,7 @@ export function registerForgeCommand(parent: Command): void {
         config?: string;
         codebase: string;
         dryRun: boolean;
+        simulate: boolean;
       }) => {
         const briefPath = path.resolve(opts.brief);
 
@@ -149,14 +205,22 @@ export function registerForgeCommand(parent: Command): void {
         const briefContent = fs.readFileSync(briefPath, 'utf-8');
         const briefClass = classifyBrief(briefContent);
         const projectRoot = opts.codebase;
+        // A real executor is never wired at CLI invocation time today, so the
+        // only executor we may construct is the explicitly-requested simulated
+        // one. Normal mode fails closed with FORGE_NO_EXECUTOR.
+        const executor = opts.simulate ? createSimulatedExecutor() : undefined;
 
         if (opts.resume) {
           const runId = opts.runId ?? generateRunId();
           const runDir = resolveRunDir(runId, projectRoot);
           console.log(`[forge] resuming run: ${runId}`);
-          const { resumePipeline } = await import('./pipeline-runner.js');
-          const result = await resumePipeline(runDir, stubExecutor);
-          console.log(`[forge] pipeline complete: ${result.runId}`);
+          try {
+            const { resumePipeline } = await import('./pipeline-runner.js');
+            const result = await resumePipeline(runDir, executor, { simulate: opts.simulate });
+            applyRunExitPolicy(result, opts.simulate);
+          } catch (err) {
+            handlePipelineError(err);
+          }
           return;
         }
 
@@ -164,7 +228,8 @@ export function registerForgeCommand(parent: Command): void {
           briefClass,
           codebase: projectRoot,
           dryRun: opts.dryRun,
-          executor: stubExecutor,
+          executor,
+          simulate: opts.simulate,
         };
 
         if (opts.dryRun) {
@@ -180,16 +245,15 @@ export function registerForgeCommand(parent: Command): void {
 
         console.log(`[forge] starting pipeline for brief: ${briefPath}`);
         console.log(`[forge] classified as: ${briefClass}`);
+        if (opts.simulate) {
+          console.log('[forge] mode: SIMULATED (explicit --simulate)');
+        }
 
         try {
           const result = await runPipeline(briefPath, projectRoot, pipelineOptions);
-          console.log(`[forge] pipeline complete: ${result.runId}`);
-          console.log(`[forge] run directory: ${result.runDir}`);
+          applyRunExitPolicy(result, opts.simulate);
         } catch (err) {
-          console.error(
-            `[forge] pipeline failed: ${err instanceof Error ? err.message : String(err)}`,
-          );
-          process.exitCode = 1;
+          handlePipelineError(err);
         }
       },
     );
@@ -224,7 +288,12 @@ export function registerForgeCommand(parent: Command): void {
     .command('resume <runId>')
     .description('Resume a stopped or failed pipeline run')
     .option('--project <path>', 'Project root (defaults to cwd)', process.cwd())
-    .action(async (runId: string, opts: { project: string }) => {
+    .option(
+      '--simulate',
+      'Simulate execution without real providers (every result is typed simulated, never verified)',
+      false,
+    )
+    .action(async (runId: string, opts: { project: string; simulate: boolean }) => {
       const runDir = resolveRunDir(runId, opts.project);
 
       if (!fs.existsSync(runDir)) {
@@ -234,15 +303,20 @@ export function registerForgeCommand(parent: Command): void {
       }
 
       console.log(`[forge] resuming run: ${runId}`);
+      if (opts.simulate) {
+        console.log('[forge] mode: SIMULATED (explicit --simulate)');
+      }
+
+      // No real executor is wired at CLI invocation time; only the explicitly
+      // requested simulated executor may be constructed (fail closed otherwise).
+      const executor = opts.simulate ? createSimulatedExecutor() : undefined;
 
       try {
         const { resumePipeline } = await import('./pipeline-runner.js');
-        const result = await resumePipeline(runDir, stubExecutor);
-        console.log(`[forge] pipeline complete: ${result.runId}`);
-        console.log(`[forge] run directory: ${result.runDir}`);
+        const result = await resumePipeline(runDir, executor, { simulate: opts.simulate });
+        applyRunExitPolicy(result, opts.simulate);
       } catch (err) {
-        console.error(`[forge] resume failed: ${err instanceof Error ? err.message : String(err)}`);
-        process.exitCode = 1;
+        handlePipelineError(err);
       }
     });
 

packages/forge/src/constants.ts

@@ -9,7 +9,16 @@ export const PACKAGE_ROOT = path.resolve(path.dirname(fileURLToPath(import.meta.
 /** Pipeline asset directory (stages, agents, rails, gates, templates). */
 export const PIPELINE_DIR = path.join(PACKAGE_ROOT, 'pipeline');
 
-/** Stage specifications — defines every pipeline stage. */
+/** Stage specifications — defines every pipeline stage.
+ *\n * Gate semantics (SDLC-D-035): every gate is one of
+ * - a real command string / GateEntry a mechanical runner can execute,
+ * - an `authority` gate (human/board sign-off; produces waiting-for-authority),
+ * - a `provider` gate (requires a wired provider such as a reviewer or CI pipeline).
+ *
+ * Vacuous gates (`true`, echo'd synthetic approvals, placeholder ci-pipeline
+ * commands) are forbidden: a stage whose gate has no real implementation
+ * fails closed instead of passing.
+ */
 export const STAGE_SPECS: Record<string, StageSpec> = {
   '00-intake': {
     number: '00',
@@ -27,7 +36,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'research',
     gate: 'discovery-complete',
     promptFile: '00b-discovery.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'discovery-complete',
+        reason: 'discovery completion is attested by an authority; no mechanical check exists',
+      },
+    ],
   },
   '01-board': {
     number: '01',
@@ -36,7 +51,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'review',
     gate: 'board-approval',
     promptFile: '01-board.md',
-    qualityGates: [{ type: 'ci-pipeline', command: 'board-approval (via board-tasks)' }],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'board-approval',
+        reason: 'board approval is a board/human decision; no mechanical gate exists',
+      },
+    ],
   },
   '01b-brief-analyzer': {
     number: '01b',
@@ -45,7 +66,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'research',
     gate: 'brief-analysis-complete',
     promptFile: '01-board.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'brief-analysis-complete',
+        reason: 'brief analysis completion is attested by an authority; no mechanical check exists',
+      },
+    ],
   },
   '02-planning-1': {
     number: '02',
@@ -54,7 +81,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'research',
     gate: 'architecture-approval',
     promptFile: '02-planning-1-architecture.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'architecture-approval',
+        reason: 'ADR approval requires authority sign-off; no mechanical check exists',
+      },
+    ],
   },
   '03-planning-2': {
     number: '03',
@@ -63,7 +96,14 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'research',
     gate: 'implementation-approval',
     promptFile: '03-planning-2-implementation.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'implementation-approval',
+        reason:
+          'implementation spec approval requires authority sign-off; no mechanical check exists',
+      },
+    ],
   },
   '04-planning-3': {
     number: '04',
@@ -72,7 +112,14 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'research',
     gate: 'decomposition-approval',
     promptFile: '04-planning-3-decomposition.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 'decomposition-approval',
+        reason:
+          'task decomposition approval requires authority sign-off; no mechanical check exists',
+      },
+    ],
   },
   '05-coding': {
     number: '05',
@@ -92,9 +139,10 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     promptFile: '06-review.md',
     qualityGates: [
       {
-        type: 'ai-review',
-        command:
-          'echo \'{"summary":"review-pass","verdict":"approve","findings":[],"stats":{"blockers":0,"should_fix":0,"suggestions":0}}\'',
+        kind: 'provider',
+        capability: 'reviewer',
+        reason:
+          'review verdicts require a wired reviewer provider; synthetic approvals are not permitted',
       },
     ],
   },
@@ -105,7 +153,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'coding',
     gate: 're-review',
     promptFile: '07-remediate.md',
-    qualityGates: ['true'],
+    qualityGates: [
+      {
+        kind: 'authority',
+        capability: 're-review',
+        reason: 'remediation re-review is an approval-based gate; no mechanical check exists',
+      },
+    ],
   },
   '08-test': {
     number: '08',
@@ -123,7 +177,13 @@ export const STAGE_SPECS: Record<string, StageSpec> = {
     type: 'deploy',
     gate: 'deploy-verification',
     promptFile: '09-deploy.md',
-    qualityGates: [{ type: 'ci-pipeline', command: 'deploy-verification' }],
+    qualityGates: [
+      {
+        kind: 'provider',
+        capability: 'ci-pipeline',
+        reason: 'deploy verification requires a wired CI pipeline provider',
+      },
+    ],
   },
 };
 

packages/forge/src/errors.ts

@@ -0,0 +1,46 @@
+/**
+ * Typed fail-closed capability errors (SDLC-D-035).
+ *
+ * A Forge run must fail closed when a required capability (executor, reviewer
+ * provider, CI pipeline, authority sign-off) is missing. These typed errors
+ * name the missing capability so callers can distinguish "not wired" from
+ * ordinary execution failures.
+ */
+
+/** Closed set of typed Forge capability error codes. */
+export const FORGE_ERROR_CODES = [
+  'FORGE_NO_EXECUTOR',
+  'FORGE_NO_REVIEWER',
+  'FORGE_NO_CI_PIPELINE',
+  'FORGE_NO_PROVIDER',
+  'FORGE_AUTHORITY_REQUIRED',
+] as const;
+
+export type ForgeErrorCode = (typeof FORGE_ERROR_CODES)[number];
+
+/** Raised when a required capability is missing and the pipeline must fail closed. */
+export class ForgeCapabilityError extends Error {
+  /** Typed error code from the closed FORGE_ERROR_CODES set. */
+  readonly code: ForgeErrorCode;
+  /** The missing capability, e.g. `task-executor`, `reviewer`, `board-approval`. */
+  readonly capability: string;
+
+  constructor(code: ForgeErrorCode, capability: string, message: string) {
+    super(message);
+    this.name = 'ForgeCapabilityError';
+    this.code = code;
+    this.capability = capability;
+  }
+}
+
+/** Map a provider gate capability to its typed error code. */
+export function providerErrorCode(capability: string): ForgeErrorCode {
+  switch (capability) {
+    case 'reviewer':
+      return 'FORGE_NO_REVIEWER';
+    case 'ci-pipeline':
+      return 'FORGE_NO_CI_PIPELINE';
+    default:
+      return 'FORGE_NO_PROVIDER';
+  }
+}

packages/forge/src/index.ts

@@ -5,6 +5,13 @@ export type {
   StageSpec,
   BriefClass,
   ClassSource,
+  ForgeOutcome,
+  AuthorityGate,
+  ProviderGate,
+  ForgeGate,
+  ForgeGateResult,
+  ForgeTaskResult,
+  RunMode,
   StageStatus,
   RunManifest,
   ForgeTaskStatus,
@@ -81,5 +88,24 @@ export {
   getPipelineStatus,
 } from './pipeline-runner.js';
 
+// Fail-closed errors and typed outcome model (SDLC-D-035)
+export { FORGE_ERROR_CODES, ForgeCapabilityError, providerErrorCode } from './errors.js';
+export type { ForgeErrorCode } from './errors.js';
+export {
+  isSatisfyingOutcome,
+  isCapabilityGate,
+  isCommandGate,
+  gateLabel,
+  uniformGateResults,
+  simulatedGateResults,
+  waitingGateResults,
+  blockedGateResults,
+  evaluateStageGates,
+} from './outcomes.js';
+export type { StageEvaluation } from './outcomes.js';
+
+// Simulated executor (explicit --simulate only)
+export { createSimulatedExecutor } from './simulated-executor.js';
+
 // CLI
 export { registerForgeCommand } from './cli.js';

packages/forge/src/outcomes.ts

@@ -0,0 +1,147 @@
+import type { GateEntry } from '@mosaicstack/macp';
+
+import type {
+  AuthorityGate,
+  ForgeGate,
+  ForgeGateResult,
+  ForgeOutcome,
+  ForgeTaskResult,
+  ProviderGate,
+} from './types.js';
+
+/**
+ * Gate and dependency satisfaction predicate (SDLC-D-035).
+ *
+ * ONLY a verified `passed` outcome satisfies. Every other member of the closed
+ * outcome set — including `simulated` — is non-satisfying, so a simulated or
+ * authority-blocked result can never be read as success-by-verification.
+ */
+export function isSatisfyingOutcome(outcome: ForgeOutcome): boolean {
+  return outcome === 'passed';
+}
+
+/** Whether a gate is an authority or provider gate (capability-based, command-less). */
+export function isCapabilityGate(gate: ForgeGate): gate is AuthorityGate | ProviderGate {
+  if (typeof gate !== 'object' || gate === null) return false;
+  const kind = (gate as Record<string, unknown>)['kind'];
+  return kind === 'authority' || kind === 'provider';
+}
+
+/** Whether a gate definition carries a real command a mechanical runner can execute. */
+export function isCommandGate(gate: ForgeGate): gate is string | GateEntry {
+  if (typeof gate === 'string') {
+    return gate.trim().length > 0;
+  }
+  if (isCapabilityGate(gate)) {
+    // Authority and provider gates are satisfied by a capability, not a command.
+    return false;
+  }
+  return typeof gate.command === 'string' && gate.command.trim().length > 0;
+}
+
+/** Typed label identifying a gate in results and logs. */
+export function gateLabel(gate: ForgeGate): string {
+  if (typeof gate === 'string') return gate;
+  if (isCapabilityGate(gate)) return `${gate.kind}:${gate.capability}`;
+  return gate.command || gate.type || 'unnamed-gate';
+}
+
+/** Reason string stamped on every simulated gate result. */
+export const SIMULATED_GATE_REASON =
+  'simulated execution (--simulate): gate was not evaluated by a real implementation';
+
+/** Build typed gate results with a uniform outcome for a stage's declared gates. */
+export function uniformGateResults(
+  gates: ForgeGate[],
+  outcome: ForgeOutcome,
+  reason: string,
+): ForgeGateResult[] {
+  return gates.map((gate) => ({ gate: gateLabel(gate), outcome, reason }));
+}
+
+/** Typed simulated gate results — used exclusively in `--simulate` runs. */
+export function simulatedGateResults(gates: ForgeGate[]): ForgeGateResult[] {
+  return uniformGateResults(gates, 'simulated', SIMULATED_GATE_REASON);
+}
+
+/** Typed waiting-for-authority gate results for approval-based stages. */
+export function waitingGateResults(gates: ForgeGate[], reason: string): ForgeGateResult[] {
+  return uniformGateResults(gates, 'waiting-for-authority', reason);
+}
+
+/** Typed blocked gate results for stages whose provider capability is not wired. */
+export function blockedGateResults(gates: ForgeGate[], reason: string): ForgeGateResult[] {
+  return uniformGateResults(gates, 'blocked', reason);
+}
+
+/** Outcome of evaluating a completed stage in normal mode. */
+export interface StageEvaluation {
+  outcome: ForgeOutcome;
+  reason: string;
+  gateResults: ForgeGateResult[];
+}
+
+/**
+ * Evaluate a stage's declared gates against the executor's typed result.
+ *
+ * Fail-closed mapping:
+ * - a `simulated` task or gate outcome in normal mode maps to `error`
+ * - a missing gate result for a required command gate maps to `blocked`
+ * - a non-passing task outcome propagates as the stage outcome
+ * - only verified `passed` task and gate outcomes yield a `passed` stage
+ */
+export function evaluateStageGates(
+  stageName: string,
+  gates: ForgeGate[],
+  result: ForgeTaskResult,
+): StageEvaluation {
+  const gateResults = result.gate_results ?? [];
+
+  if (result.outcome === 'simulated') {
+    return {
+      outcome: 'error',
+      reason: `executor reported a simulated outcome for stage '${stageName}' in normal mode — refusing to treat simulated results as verified`,
+      gateResults,
+    };
+  }
+
+  if (!isSatisfyingOutcome(result.outcome)) {
+    return {
+      outcome: result.outcome,
+      reason: `task outcome is '${result.outcome}': ${result.reason}`,
+      gateResults,
+    };
+  }
+
+  for (const gate of gates) {
+    // Authority and provider gates are pre-flighted before execution; they have
+    // no mechanical result to verify here.
+    if (!isCommandGate(gate)) continue;
+
+    const label = gateLabel(gate);
+    const gateResult = gateResults.find((r) => r.gate === label);
+    if (!gateResult) {
+      return {
+        outcome: 'blocked',
+        reason: `no gate result was reported for required gate '${label}' (stage '${stageName}')`,
+        gateResults,
+      };
+    }
+    if (!isSatisfyingOutcome(gateResult.outcome)) {
+      return {
+        outcome: gateResult.outcome === 'simulated' ? 'error' : gateResult.outcome,
+        reason: `gate '${label}' outcome is '${gateResult.outcome}': ${gateResult.reason}`,
+        gateResults,
+      };
+    }
+  }
+
+  return {
+    outcome: 'passed',
+    reason:
+      gates.length === 0
+        ? "stage declares no gates; task outcome 'passed' accepted"
+        : 'all declared gates verified passed',
+    gateResults,
+  };
+}

packages/forge/src/pipeline-runner.ts

@@ -1,18 +1,33 @@
 import fs from 'node:fs';
 import path from 'node:path';
 
-import { STAGE_SEQUENCE } from './constants.js';
+import { STAGE_SEQUENCE, STAGE_SPECS } from './constants.js';
 import { determineBriefClass, stagesForClass } from './brief-classifier.js';
+import { ForgeCapabilityError, providerErrorCode } from './errors.js';
+import {
+  blockedGateResults,
+  evaluateStageGates,
+  isCapabilityGate,
+  simulatedGateResults,
+  waitingGateResults,
+} from './outcomes.js';
 import { mapStageToTask } from './stage-adapter.js';
+import { createSimulatedExecutor } from './simulated-executor.js';
 import type {
   ForgeTask,
+  ForgeTaskResult,
   PipelineOptions,
   PipelineResult,
   RunManifest,
+  RunMode,
   StageStatus,
   TaskExecutor,
 } from './types.js';
 
+/** Reason stamped on stages that complete under explicit simulation. */
+const SIMULATED_STAGE_REASON =
+  'simulated execution (--simulate): stage was not executed by a real executor';
+
 /**
  * Generate a timestamp-based run ID.
  */
@@ -47,6 +62,7 @@ function createManifest(opts: {
   briefClass: RunManifest['briefClass'];
   classSource: RunManifest['classSource'];
   forceBoard: boolean;
+  mode: RunMode;
   runDir: string;
 }): RunManifest {
   const ts = nowISO();
@@ -57,6 +73,7 @@ function createManifest(opts: {
     briefClass: opts.briefClass,
     classSource: opts.classSource,
     forceBoard: opts.forceBoard,
+    mode: opts.mode,
     createdAt: ts,
     updatedAt: ts,
     currentStage: '',
@@ -108,20 +125,199 @@ export function selectStages(stages?: string[], skipTo?: string): string[] {
   return selected.slice(skipIndex);
 }
 
+/**
+ * Fail closed when the required executor capability is missing (SDLC-D-035).
+ */
+function requireExecutor(executor: TaskExecutor | undefined, simulate: boolean): TaskExecutor {
+  if (executor) return executor;
+  if (simulate) return createSimulatedExecutor({ log: false });
+  throw new ForgeCapabilityError(
+    'FORGE_NO_EXECUTOR',
+    'task-executor',
+    'no task executor is wired; refusing to run the pipeline with a stub executor (fail closed). ' +
+      'Pass --simulate to opt into explicitly simulated execution.',
+  );
+}
+
+/**
+ * Pre-flight a stage's gates in normal mode (fail closed, SDLC-D-035).
+ *
+ * - authority gates: record a typed `waiting-for-authority` stage result and
+ *   raise FORGE_AUTHORITY_REQUIRED — approval-based gates never pass vacuously.
+ * - provider gates: record a typed `blocked` stage result and raise the typed
+ *   capability error for the missing provider.
+ *
+ * Returns the stage status to record when the pre-flight blocks, or undefined
+ * when the stage may proceed.
+ */
+function preflightStageGates(
+  stageName: string,
+  manifest: RunManifest,
+): { status: StageStatus; error: ForgeCapabilityError } | undefined {
+  const spec = STAGE_SPECS[stageName];
+  if (!spec) throw new Error(`Unknown Forge stage: ${stageName}`);
+
+  for (const gate of spec.qualityGates) {
+    if (!isCapabilityGate(gate)) continue;
+
+    const startedAt = manifest.stages[stageName]?.startedAt;
+    const completedAt = nowISO();
+
+    if (gate.kind === 'authority') {
+      const reason = `gate '${gate.capability}' requires authority sign-off; no mechanical implementation exists (${gate.reason})`;
+      return {
+        status: {
+          status: 'waiting-for-authority',
+          reason,
+          startedAt,
+          completedAt,
+          gateResults: waitingGateResults(spec.qualityGates, reason),
+        },
+        error: new ForgeCapabilityError(
+          'FORGE_AUTHORITY_REQUIRED',
+          gate.capability,
+          `stage '${stageName}' is blocked on authority gate '${gate.capability}': ${gate.reason}. ` +
+            'The pipeline fails closed instead of passing vacuously. Record the approval out-of-band ' +
+            'or run with --simulate for explicitly simulated execution.',
+        ),
+      };
+    }
+
+    const reason = `gate '${gate.capability}' requires provider '${gate.capability}' and none is wired (${gate.reason})`;
+    return {
+      status: {
+        status: 'blocked',
+        reason,
+        startedAt,
+        completedAt,
+        gateResults: blockedGateResults(spec.qualityGates, reason),
+      },
+      error: new ForgeCapabilityError(
+        providerErrorCode(gate.capability),
+        gate.capability,
+        `stage '${stageName}' requires provider '${gate.capability}' which is not wired: ${gate.reason}. ` +
+          'The pipeline fails closed instead of passing vacuously.',
+      ),
+    };
+  }
+
+  return undefined;
+}
+
+/**
+ * Execute the given stage tasks sequentially, updating the manifest.
+ *
+ * Normal mode requires a real executor and evaluates every declared command
+ * gate through the typed outcome model; any non-verified result fails closed.
+ * Simulate mode types every stage and gate result as `simulated`.
+ */
+async function executeStages(opts: {
+  manifest: RunManifest;
+  runDir: string;
+  tasks: ForgeTask[];
+  stageNames: string[];
+  executor: TaskExecutor;
+  simulate: boolean;
+}): Promise<void> {
+  const { manifest, runDir, tasks, stageNames, executor, simulate } = opts;
+
+  for (let i = 0; i < tasks.length; i++) {
+    const task = tasks[i]!;
+    const stageName = stageNames[i]!;
+    const spec = STAGE_SPECS[stageName];
+    if (!spec) throw new Error(`Unknown Forge stage: ${stageName}`);
+
+    // Update manifest: stage in progress
+    manifest.currentStage = stageName;
+    manifest.stages[stageName] = {
+      status: 'in_progress',
+      startedAt: nowISO(),
+    };
+    saveManifest(runDir, manifest);
+
+    // Fail-closed pre-flight (normal mode only): authority/provider gates have
+    // no mechanical implementation and must never pass vacuously.
+    if (!simulate) {
+      const blocked = preflightStageGates(stageName, manifest);
+      if (blocked) {
+        manifest.stages[stageName] = blocked.status;
+        manifest.status =
+          blocked.status.status === 'waiting-for-authority' ? 'waiting-for-authority' : 'failed';
+        saveManifest(runDir, manifest);
+        throw blocked.error;
+      }
+    }
+
+    let result: ForgeTaskResult;
+    try {
+      await executor.submitTask(task);
+      result = await executor.waitForCompletion(task.id, task.timeoutSeconds * 1000);
+    } catch (error) {
+      // Process errors (including timeouts) map to the fail-closed `error` outcome.
+      const reason = error instanceof Error ? error.message : String(error);
+      manifest.stages[stageName] = {
+        status: 'error',
+        reason: `executor error: ${reason}`,
+        startedAt: manifest.stages[stageName]?.startedAt,
+        completedAt: nowISO(),
+        gateResults: [],
+      };
+      manifest.status = 'failed';
+      saveManifest(runDir, manifest);
+      throw error instanceof Error ? error : new Error(reason);
+    }
+
+    if (simulate) {
+      manifest.stages[stageName] = {
+        status: 'simulated',
+        reason: SIMULATED_STAGE_REASON,
+        startedAt: manifest.stages[stageName]?.startedAt,
+        completedAt: nowISO(),
+        gateResults: simulatedGateResults(spec.qualityGates),
+      };
+      saveManifest(runDir, manifest);
+      continue;
+    }
+
+    const evaluation = evaluateStageGates(stageName, spec.qualityGates, result);
+    manifest.stages[stageName] = {
+      status: evaluation.outcome,
+      reason: evaluation.reason,
+      startedAt: manifest.stages[stageName]?.startedAt,
+      completedAt: nowISO(),
+      gateResults: evaluation.gateResults,
+    };
+
+    if (evaluation.outcome !== 'passed') {
+      manifest.status =
+        evaluation.outcome === 'waiting-for-authority' ? 'waiting-for-authority' : 'failed';
+      saveManifest(runDir, manifest);
+      throw new Error(`Stage ${stageName} ${evaluation.outcome}: ${evaluation.reason}`);
+    }
+
+    saveManifest(runDir, manifest);
+  }
+}
+
 /**
  * Run the Forge pipeline.
  *
- * 1. Classify the brief
- * 2. Generate a run ID and create run directory
- * 3. Map stages to tasks and submit to TaskExecutor
- * 4. Track manifest with stage statuses
- * 5. Return pipeline result
+ * 1. Fail closed unless a real executor is wired or simulation is explicit
+ * 2. Classify the brief
+ * 3. Generate a run ID and create run directory
+ * 4. Map stages to tasks and submit to TaskExecutor
+ * 5. Track manifest with typed stage outcomes
+ * 6. Return pipeline result
  */
 export async function runPipeline(
   briefPath: string,
   projectRoot: string,
   options: PipelineOptions,
 ): Promise<PipelineResult> {
+  const simulate = options.simulate ?? false;
+  const executor = requireExecutor(options.executor, simulate);
+  const mode: RunMode = simulate ? 'simulated' : 'normal';
+
   const resolvedRoot = path.resolve(projectRoot);
   const resolvedBrief = path.resolve(briefPath);
   const briefContent = fs.readFileSync(resolvedBrief, 'utf-8');
@@ -146,6 +342,7 @@ export async function runPipeline(
     briefClass,
     classSource,
     forceBoard: options.forceBoard ?? false,
+    mode,
     runDir,
   });
 
@@ -172,54 +369,10 @@ export async function runPipeline(
   }
 
   // Execute stages
-  const { executor } = options;
-  for (let i = 0; i < tasks.length; i++) {
-    const task = tasks[i]!;
-    const stageName = selectedStages[i]!;
+  await executeStages({ manifest, runDir, tasks, stageNames: selectedStages, executor, simulate });
 
-    // Update manifest: stage in progress
-    manifest.currentStage = stageName;
-    manifest.stages[stageName] = {
-      status: 'in_progress',
-      startedAt: nowISO(),
-    };
-    saveManifest(runDir, manifest);
-
-    try {
-      await executor.submitTask(task);
-      const result = await executor.waitForCompletion(task.id, task.timeoutSeconds * 1000);
-
-      // Update manifest: stage completed or failed
-      const stageStatus: StageStatus = {
-        status: result.status === 'completed' ? 'passed' : 'failed',
-        startedAt: manifest.stages[stageName]!.startedAt,
-        completedAt: nowISO(),
-      };
-      manifest.stages[stageName] = stageStatus;
-
-      if (result.status !== 'completed') {
-        manifest.status = 'failed';
-        saveManifest(runDir, manifest);
-        throw new Error(`Stage ${stageName} failed with status: ${result.status}`);
-      }
-
-      saveManifest(runDir, manifest);
-    } catch (error) {
-      if (!manifest.stages[stageName]?.completedAt) {
-        manifest.stages[stageName] = {
-          status: 'failed',
-          startedAt: manifest.stages[stageName]?.startedAt,
-          completedAt: nowISO(),
-        };
-      }
-      manifest.status = 'failed';
-      saveManifest(runDir, manifest);
-      throw error;
-    }
-  }
-
-  // All stages passed
-  manifest.status = 'completed';
+  // All stages reached a terminal state for this mode
+  manifest.status = simulate ? 'simulated' : 'completed';
   saveManifest(runDir, manifest);
 
   return {
@@ -234,22 +387,30 @@ export async function runPipeline(
 }
 
 /**
- * Resume a pipeline from the last incomplete stage.
+ * Resume a pipeline from the last non-passed stage.
  */
 export async function resumePipeline(
   runDir: string,
-  executor: TaskExecutor,
+  executor?: TaskExecutor,
+  options?: { simulate?: boolean },
 ): Promise<PipelineResult> {
+  const simulate = options?.simulate ?? false;
+  const wiredExecutor = requireExecutor(executor, simulate);
+  const mode: RunMode = simulate ? 'simulated' : 'normal';
+
   const manifest = loadManifest(runDir);
   const resolvedRoot = path.dirname(path.dirname(path.dirname(runDir))); // .forge/runs/{id} → project root
 
   const briefContent = fs.readFileSync(manifest.brief, 'utf-8');
   const allStages = stagesForClass(manifest.briefClass, manifest.forceBoard);
 
-  // Find first non-passed stage
+  manifest.mode = mode;
+
+  // Find first non-satisfying stage (only a verified `passed` counts as done;
+  // simulated and waiting-for-authority stages are re-run).
   const resumeFrom = allStages.find((s) => manifest.stages[s]?.status !== 'passed');
   if (!resumeFrom) {
-    manifest.status = 'completed';
+    manifest.status = mode === 'simulated' ? 'simulated' : 'completed';
     saveManifest(runDir, manifest);
     return {
       runId: manifest.runId,
@@ -284,49 +445,16 @@ export async function resumePipeline(
     tasks.push(task);
   }
 
-  for (let i = 0; i < tasks.length; i++) {
-    const task = tasks[i]!;
-    const stageName = remainingStages[i]!;
+  await executeStages({
+    manifest,
+    runDir,
+    tasks,
+    stageNames: remainingStages,
+    executor: wiredExecutor,
+    simulate,
+  });
 
-    manifest.currentStage = stageName;
-    manifest.stages[stageName] = {
-      status: 'in_progress',
-      startedAt: nowISO(),
-    };
-    saveManifest(runDir, manifest);
-
-    try {
-      await executor.submitTask(task);
-      const result = await executor.waitForCompletion(task.id, task.timeoutSeconds * 1000);
-
-      manifest.stages[stageName] = {
-        status: result.status === 'completed' ? 'passed' : 'failed',
-        startedAt: manifest.stages[stageName]!.startedAt,
-        completedAt: nowISO(),
-      };
-
-      if (result.status !== 'completed') {
-        manifest.status = 'failed';
-        saveManifest(runDir, manifest);
-        throw new Error(`Stage ${stageName} failed with status: ${result.status}`);
-      }
-
-      saveManifest(runDir, manifest);
-    } catch (error) {
-      if (!manifest.stages[stageName]?.completedAt) {
-        manifest.stages[stageName] = {
-          status: 'failed',
-          startedAt: manifest.stages[stageName]?.startedAt,
-          completedAt: nowISO(),
-        };
-      }
-      manifest.status = 'failed';
-      saveManifest(runDir, manifest);
-      throw error;
-    }
-  }
-
-  manifest.status = 'completed';
+  manifest.status = simulate ? 'simulated' : 'completed';
   saveManifest(runDir, manifest);
 
   return {

packages/forge/src/simulated-executor.ts

@@ -0,0 +1,32 @@
+import type { ForgeTask, ForgeTaskResult, TaskExecutor } from './types.js';
+
+/**
+ * Simulated executor — used ONLY when the caller explicitly passes --simulate.
+ *
+ * It submits no real work and returns typed `simulated` results so a simulated
+ * run can never be confused with a verified one. In normal mode (no --simulate)
+ * the CLI refuses to run at all with FORGE_NO_EXECUTOR instead of wiring this
+ * stub (fail closed, SDLC-D-035).
+ */
+export function createSimulatedExecutor(options?: { log?: boolean }): TaskExecutor {
+  const log = options?.log ?? true;
+  return {
+    async submitTask(task: ForgeTask) {
+      if (log) console.log(`  [forge:simulated] stage submitted: ${task.id} (${task.title})`);
+    },
+    async waitForCompletion(taskId: string): Promise<ForgeTaskResult> {
+      if (log) console.log(`  [forge:simulated] stage complete:  ${taskId}`);
+      return {
+        task_id: taskId,
+        outcome: 'simulated',
+        reason: 'no executor wired; simulated execution requested via --simulate',
+        completed_at: new Date().toISOString(),
+        exit_code: 0,
+        gate_results: [],
+      };
+    },
+    async getTaskStatus() {
+      return 'completed' as const;
+    },
+  };
+}

packages/forge/src/types.ts

@@ -1,4 +1,4 @@
-import type { GateEntry, TaskResult } from '@mosaicstack/macp';
+import type { GateEntry } from '@mosaicstack/macp';
 
 /** Stage dispatch mode. */
 export type StageDispatch = 'exec' | 'yolo' | 'pi';
@@ -6,6 +6,58 @@ export type StageDispatch = 'exec' | 'yolo' | 'pi';
 /** Stage type — determines agent selection and gate requirements. */
 export type StageType = 'research' | 'review' | 'coding' | 'deploy';
 
+/**
+ * Typed outcome for every gate and stage evaluation — closed set (SDLC-D-035).
+ *
+ * Only `passed` means "verified by a real implementation". `simulated` is
+ * produced exclusively in explicit `--simulate` runs and is never satisfying.
+ */
+export type ForgeOutcome =
+  | 'passed'
+  | 'failed'
+  | 'blocked'
+  | 'error'
+  | 'waiting-for-authority'
+  | 'simulated'
+  | 'not-applicable';
+
+/** A gate that requires authority (human/board) sign-off; no mechanical command can satisfy it. */
+export interface AuthorityGate {
+  kind: 'authority';
+  capability: string;
+  reason: string;
+}
+
+/** A gate that requires a wired provider (e.g. an AI reviewer, CI pipeline) to evaluate. */
+export interface ProviderGate {
+  kind: 'provider';
+  capability: string;
+  reason: string;
+}
+
+/** Forge quality gate: a real command, an authority sign-off, or a provider-backed check. */
+export type ForgeGate = string | GateEntry | AuthorityGate | ProviderGate;
+
+/** Typed result of evaluating a single quality gate. */
+export interface ForgeGateResult {
+  gate: string;
+  outcome: ForgeOutcome;
+  reason: string;
+  exitCode?: number;
+  output?: string;
+  timedOut?: boolean;
+}
+
+/** Typed result of a task/stage execution returned by a TaskExecutor. */
+export interface ForgeTaskResult {
+  task_id: string;
+  outcome: ForgeOutcome;
+  reason: string;
+  completed_at: string;
+  exit_code: number;
+  gate_results: ForgeGateResult[];
+}
+
 /** Stage specification — defines a single pipeline stage. */
 export interface StageSpec {
   number: string;
@@ -14,7 +66,7 @@ export interface StageSpec {
   type: StageType;
   gate: string;
   promptFile: string;
-  qualityGates: (string | GateEntry)[];
+  qualityGates: ForgeGate[];
 }
 
 /** Brief classification. */
@@ -25,11 +77,18 @@ export type ClassSource = 'cli' | 'frontmatter' | 'auto';
 
 /** Per-stage status within a run manifest. */
 export interface StageStatus {
-  status: 'pending' | 'in_progress' | 'passed' | 'failed';
+  status: 'pending' | 'in_progress' | ForgeOutcome;
+  /** Why the stage reached its current (terminal) outcome, when applicable. */
+  reason?: string;
   startedAt?: string;
   completedAt?: string;
+  /** Typed per-gate results recorded alongside the stage outcome. */
+  gateResults?: ForgeGateResult[];
 }
 
+/** Execution mode of a run. */
+export type RunMode = 'normal' | 'simulated';
+
 /** Run manifest — persisted to disk as manifest.json. */
 export interface RunManifest {
   runId: string;
@@ -38,10 +97,23 @@ export interface RunManifest {
   briefClass: BriefClass;
   classSource: ClassSource;
   forceBoard: boolean;
+  /**
+   * Execution mode. `simulated` runs stub execution; their results are typed
+   * `simulated` and must never be read as verified success. Optional because
+   * manifests written before this field existed default to `normal`.
+   */
+  mode?: RunMode;
   createdAt: string;
   updatedAt: string;
   currentStage: string;
-  status: 'in_progress' | 'completed' | 'failed' | 'interrupted' | 'rejected';
+  status:
+    | 'in_progress'
+    | 'completed'
+    | 'failed'
+    | 'interrupted'
+    | 'rejected'
+    | 'simulated'
+    | 'waiting-for-authority';
   stages: Record<string, StageStatus>;
 }
 
@@ -65,7 +137,7 @@ export interface ForgeTask {
   briefPath: string;
   resultPath: string;
   timeoutSeconds: number;
-  qualityGates: (string | GateEntry)[];
+  qualityGates: ForgeGate[];
   worktree?: string;
   command?: string;
   dependsOn?: string[];
@@ -76,7 +148,7 @@ export interface ForgeTask {
 /** Abstract task executor — decouples from packages/coord. */
 export interface TaskExecutor {
   submitTask(task: ForgeTask): Promise<void>;
-  waitForCompletion(taskId: string, timeoutMs: number): Promise<TaskResult>;
+  waitForCompletion(taskId: string, timeoutMs: number): Promise<ForgeTaskResult>;
   getTaskStatus(taskId: string): Promise<ForgeTaskStatus>;
 }
 
@@ -122,7 +194,16 @@ export interface PipelineOptions {
   stages?: string[];
   skipTo?: string;
   dryRun?: boolean;
-  executor: TaskExecutor;
+  /**
+   * Real task executor. Required in normal mode: the pipeline fails closed
+   * with FORGE_NO_EXECUTOR when it is absent.
+   */
+  executor?: TaskExecutor;
+  /**
+   * Explicit opt-in to simulated execution. Every stage and gate result is
+   * typed `simulated` and is never satisfying.
+   */
+  simulate?: boolean;
 }
 
 /** Pipeline run result. */

packages/macp/__tests__/gate-runner.test.ts

@@ -1,253 +0,0 @@
-import { mkdirSync, readFileSync, rmSync } from 'node:fs';
-import { join } from 'node:path';
-import { tmpdir } from 'node:os';
-import { randomUUID } from 'node:crypto';
-import { describe, it, expect, beforeEach, afterEach } from 'vitest';
-import { normalizeGate, countAIFindings, runGate, runGates } from '../src/gate-runner.js';
-
-function makeTmpDir(): string {
-  const dir = join(tmpdir(), `macp-gate-${randomUUID()}`);
-  mkdirSync(dir, { recursive: true });
-  return dir;
-}
-
-describe('normalizeGate', () => {
-  it('normalizes a string to mechanical gate', () => {
-    expect(normalizeGate('echo test')).toEqual({
-      command: 'echo test',
-      type: 'mechanical',
-      fail_on: 'blocker',
-    });
-  });
-
-  it('normalizes an object gate with defaults', () => {
-    expect(normalizeGate({ command: 'lint' })).toEqual({
-      command: 'lint',
-      type: 'mechanical',
-      fail_on: 'blocker',
-    });
-  });
-
-  it('preserves explicit type and fail_on', () => {
-    expect(normalizeGate({ command: 'review', type: 'ai-review', fail_on: 'any' })).toEqual({
-      command: 'review',
-      type: 'ai-review',
-      fail_on: 'any',
-    });
-  });
-
-  it('handles non-string/non-object input', () => {
-    expect(normalizeGate(42)).toEqual({ command: '', type: 'mechanical', fail_on: 'blocker' });
-    expect(normalizeGate(null)).toEqual({ command: '', type: 'mechanical', fail_on: 'blocker' });
-  });
-});
-
-describe('countAIFindings', () => {
-  it('returns zeros for non-object', () => {
-    expect(countAIFindings(null)).toEqual({ blockers: 0, total: 0 });
-    expect(countAIFindings('string')).toEqual({ blockers: 0, total: 0 });
-    expect(countAIFindings([])).toEqual({ blockers: 0, total: 0 });
-  });
-
-  it('counts from stats block', () => {
-    const output = { stats: { blockers: 2, should_fix: 3, suggestions: 1 } };
-    expect(countAIFindings(output)).toEqual({ blockers: 2, total: 6 });
-  });
-
-  it('counts from findings array when stats has no blockers', () => {
-    const output = {
-      stats: { blockers: 0 },
-      findings: [{ severity: 'blocker' }, { severity: 'warning' }, { severity: 'blocker' }],
-    };
-    expect(countAIFindings(output)).toEqual({ blockers: 2, total: 3 });
-  });
-
-  it('uses stats blockers over findings array when stats has blockers', () => {
-    const output = {
-      stats: { blockers: 5 },
-      findings: [{ severity: 'blocker' }, { severity: 'warning' }],
-    };
-    // stats.blockers = 5, total from stats = 5+0+0 = 5, findings not used for total since stats total is non-zero
-    expect(countAIFindings(output)).toEqual({ blockers: 5, total: 5 });
-  });
-
-  it('counts findings length as total when stats has zero total', () => {
-    const output = {
-      findings: [{ severity: 'warning' }, { severity: 'info' }],
-    };
-    expect(countAIFindings(output)).toEqual({ blockers: 0, total: 2 });
-  });
-});
-
-describe('runGate', () => {
-  let tmp: string;
-  let logPath: string;
-
-  beforeEach(() => {
-    tmp = makeTmpDir();
-    logPath = join(tmp, 'gate.log');
-  });
-
-  afterEach(() => {
-    rmSync(tmp, { recursive: true, force: true });
-  });
-
-  it('passes mechanical gate on exit 0', () => {
-    const result = runGate('echo hello', tmp, logPath, 30);
-    expect(result.passed).toBe(true);
-    expect(result.exit_code).toBe(0);
-    expect(result.type).toBe('mechanical');
-    expect(result.output).toContain('hello');
-  });
-
-  it('fails mechanical gate on non-zero exit', () => {
-    const result = runGate('exit 1', tmp, logPath, 30);
-    expect(result.passed).toBe(false);
-    expect(result.exit_code).toBe(1);
-  });
-
-  it('ci-pipeline always passes', () => {
-    const result = runGate({ command: 'anything', type: 'ci-pipeline' }, tmp, logPath, 30);
-    expect(result.passed).toBe(true);
-    expect(result.type).toBe('ci-pipeline');
-    expect(result.output).toBe('CI pipeline gate placeholder');
-  });
-
-  it('empty command passes', () => {
-    const result = runGate({ command: '' }, tmp, logPath, 30);
-    expect(result.passed).toBe(true);
-  });
-
-  it('ai-review gate parses JSON output', () => {
-    const json = JSON.stringify({ stats: { blockers: 0, should_fix: 1 } });
-    const result = runGate({ command: `echo '${json}'`, type: 'ai-review' }, tmp, logPath, 30);
-    expect(result.passed).toBe(true);
-    expect(result.blockers).toBe(0);
-    expect(result.findings).toBe(1);
-  });
-
-  it('ai-review gate fails on blockers', () => {
-    const json = JSON.stringify({ stats: { blockers: 2 } });
-    const result = runGate({ command: `echo '${json}'`, type: 'ai-review' }, tmp, logPath, 30);
-    expect(result.passed).toBe(false);
-    expect(result.blockers).toBe(2);
-  });
-
-  it('ai-review gate with fail_on=any fails on any findings', () => {
-    const json = JSON.stringify({ stats: { blockers: 0, should_fix: 1 } });
-    const result = runGate(
-      { command: `echo '${json}'`, type: 'ai-review', fail_on: 'any' },
-      tmp,
-      logPath,
-      30,
-    );
-    expect(result.passed).toBe(false);
-    expect(result.fail_on).toBe('any');
-  });
-
-  it('ai-review gate fails on invalid JSON output', () => {
-    const result = runGate({ command: 'echo "not json"', type: 'ai-review' }, tmp, logPath, 30);
-    expect(result.passed).toBe(false);
-    expect(result.parse_error).toBeDefined();
-  });
-
-  it('writes to log file', () => {
-    runGate('echo logged', tmp, logPath, 30);
-    const log = readFileSync(logPath, 'utf-8');
-    expect(log).toContain('COMMAND: echo logged');
-    expect(log).toContain('logged');
-    expect(log).toContain('EXIT:');
-  });
-});
-
-describe('runGates', () => {
-  let tmp: string;
-  let logPath: string;
-  let eventsPath: string;
-
-  beforeEach(() => {
-    tmp = makeTmpDir();
-    logPath = join(tmp, 'gates.log');
-    eventsPath = join(tmp, 'events.ndjson');
-  });
-
-  afterEach(() => {
-    rmSync(tmp, { recursive: true, force: true });
-  });
-
-  it('runs multiple gates and returns results', () => {
-    const { allPassed, gateResults } = runGates(
-      ['echo one', 'echo two'],
-      tmp,
-      logPath,
-      30,
-      eventsPath,
-      'task-1',
-    );
-    expect(allPassed).toBe(true);
-    expect(gateResults).toHaveLength(2);
-  });
-
-  it('reports failure when any gate fails', () => {
-    const { allPassed, gateResults } = runGates(
-      ['echo ok', 'exit 1'],
-      tmp,
-      logPath,
-      30,
-      eventsPath,
-      'task-2',
-    );
-    expect(allPassed).toBe(false);
-    expect(gateResults[0]!.passed).toBe(true);
-    expect(gateResults[1]!.passed).toBe(false);
-  });
-
-  it('emits events for each gate', () => {
-    runGates(['echo test'], tmp, logPath, 30, eventsPath, 'task-3');
-    const events = readFileSync(eventsPath, 'utf-8')
-      .trim()
-      .split('\n')
-      .map((l) => JSON.parse(l));
-    expect(events).toHaveLength(2); // started + passed
-    expect(events[0].event_type).toBe('rail.check.started');
-    expect(events[1].event_type).toBe('rail.check.passed');
-  });
-
-  it('skips gates with empty command (non ci-pipeline)', () => {
-    const { gateResults } = runGates(
-      [{ command: '', type: 'mechanical' }, 'echo real'],
-      tmp,
-      logPath,
-      30,
-      eventsPath,
-      'task-4',
-    );
-    expect(gateResults).toHaveLength(1);
-  });
-
-  it('does not skip ci-pipeline even with empty command', () => {
-    const { gateResults } = runGates(
-      [{ command: '', type: 'ci-pipeline' }],
-      tmp,
-      logPath,
-      30,
-      eventsPath,
-      'task-5',
-    );
-    expect(gateResults).toHaveLength(1);
-    expect(gateResults[0]!.passed).toBe(true);
-  });
-
-  it('emits failed event with correct message', () => {
-    runGates(['exit 42'], tmp, logPath, 30, eventsPath, 'task-6');
-    const events = readFileSync(eventsPath, 'utf-8')
-      .trim()
-      .split('\n')
-      .map((l) => JSON.parse(l));
-    const failEvent = events.find(
-      (e: Record<string, unknown>) => e.event_type === 'rail.check.failed',
-    );
-    expect(failEvent).toBeDefined();
-    expect(failEvent.message).toContain('Gate failed (');
-  });
-});

packages/macp/src/cli.spec.ts

@@ -1,5 +1,8 @@
-import { describe, it, expect } from 'vitest';
+import { describe, it, expect, afterEach, beforeEach, vi } from 'vitest';
 import { Command } from 'commander';
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
 import { registerMacpCommand } from './cli.js';
 
 describe('registerMacpCommand', () => {
@@ -75,3 +78,162 @@ describe('registerMacpCommand', () => {
     expect(topLevel).toContain('events');
   });
 });
+
+/**
+ * RI-N2 fail-closed CLI behavior: an unimplemented capability is a failure,
+ * never a success. Every stub exits nonzero with a typed message, and the
+ * implemented `macp gate` mirrors the typed gate-runner states.
+ */
+describe('registerMacpCommand fail-closed (RI-N2)', () => {
+  let tmpDir: string;
+
+  function buildProgram(): Command {
+    const program = new Command();
+    program.exitOverride();
+    program.configureOutput({ writeErr: () => {} });
+    registerMacpCommand(program);
+    return program;
+  }
+
+  beforeEach(() => {
+    tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), 'macp-cli-failclosed-'));
+    process.exitCode = 0;
+  });
+
+  afterEach(() => {
+    process.exitCode = 0;
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  it('macp tasks list exits nonzero (unimplemented capability)', async () => {
+    const program = buildProgram();
+    await program.parseAsync(['macp', 'tasks', 'list'], { from: 'user' });
+    expect(process.exitCode).not.toBe(0);
+  });
+
+  it('macp submit exits nonzero with a typed MACP_NOT_IMPLEMENTED message', async () => {
+    const program = buildProgram();
+    const errSpy = vi.spyOn(console, 'error').mockImplementation(() => {});
+    try {
+      await program.parseAsync(['macp', 'submit', 'spec.json'], { from: 'user' });
+      expect(process.exitCode).not.toBe(0);
+      const errText = errSpy.mock.calls.map((c) => String(c[0])).join('\n');
+      expect(errText).toContain('MACP_NOT_IMPLEMENTED');
+    } finally {
+      errSpy.mockRestore();
+    }
+  });
+
+  it('macp events tail exits nonzero (unimplemented capability)', async () => {
+    const program = buildProgram();
+    await program.parseAsync(['macp', 'events', 'tail'], { from: 'user' });
+    expect(process.exitCode).not.toBe(0);
+  });
+
+  it('macp gate runs a green inline command and exits 0', async () => {
+    const program = buildProgram();
+    await program.parseAsync(
+      [
+        'macp',
+        'gate',
+        'exit 0',
+        '--cwd',
+        tmpDir,
+        '--log',
+        path.join(tmpDir, 'g.log'),
+        '--timeout',
+        '10',
+      ],
+      { from: 'user' },
+    );
+    expect(process.exitCode).toBe(0);
+  });
+
+  it('macp gate exits nonzero on a failing command', async () => {
+    const program = buildProgram();
+    await program.parseAsync(
+      [
+        'macp',
+        'gate',
+        'exit 9',
+        '--cwd',
+        tmpDir,
+        '--log',
+        path.join(tmpDir, 'g.log'),
+        '--timeout',
+        '10',
+      ],
+      { from: 'user' },
+    );
+    expect(process.exitCode).not.toBe(0);
+  });
+
+  it('macp gate with an unimplemented ci-pipeline capability exits nonzero', async () => {
+    const program = buildProgram();
+    const specPath = path.join(tmpDir, 'gates.json');
+    fs.writeFileSync(specPath, JSON.stringify([{ type: 'ci-pipeline' }]));
+    await program.parseAsync(
+      [
+        'macp',
+        'gate',
+        specPath,
+        '--cwd',
+        tmpDir,
+        '--log',
+        path.join(tmpDir, 'g.log'),
+        '--timeout',
+        '10',
+      ],
+      { from: 'user' },
+    );
+    expect(process.exitCode).not.toBe(0);
+  });
+
+  it('macp gate --simulate completes (exit 0) but reports simulated results', async () => {
+    const program = buildProgram();
+    const logSpy = vi.spyOn(console, 'log').mockImplementation(() => {});
+    try {
+      await program.parseAsync(
+        [
+          'macp',
+          'gate',
+          'exit 0',
+          '--simulate',
+          '--cwd',
+          tmpDir,
+          '--log',
+          path.join(tmpDir, 'g.log'),
+          '--timeout',
+          '10',
+        ],
+        { from: 'user' },
+      );
+      // completes only because the caller explicitly asked to simulate
+      expect(process.exitCode).toBe(0);
+      const outText = logSpy.mock.calls.map((c) => String(c[0])).join('\n');
+      expect(outText).toContain('simulated');
+      expect(outText).toContain('SIMULATED');
+    } finally {
+      logSpy.mockRestore();
+    }
+  });
+
+  it('macp gate with an empty spec exits nonzero with a typed error', async () => {
+    const program = buildProgram();
+    await program.parseAsync(
+      [
+        'macp',
+        'gate',
+        '   ',
+        '--cwd',
+        tmpDir,
+        '--log',
+        path.join(tmpDir, 'g.log'),
+        '--timeout',
+        '10',
+      ],
+      { from: 'user' },
+    );
+    expect(process.exitCode).not.toBe(0);
+  });
+});

packages/macp/src/cli.ts

@@ -1,5 +1,73 @@
+import { existsSync, readFileSync } from 'node:fs';
+
 import type { Command } from 'commander';
 
+import { runGates } from './gate-runner.js';
+import { MACPCapabilityError, type MacpErrorCode } from './errors.js';
+
+/**
+ * Load gates from a spec: an existing file (JSON gates array, a JSON object
+ * with `quality_gates`, a JSON gate object, or one command per line) or an
+ * inline command string. Fails closed with a typed capability error when the
+ * spec contains no executable gate definition.
+ */
+function loadGateSpec(spec: string): unknown[] {
+  if (existsSync(spec)) {
+    const raw = readFileSync(spec, 'utf-8');
+    try {
+      const parsed = JSON.parse(raw) as unknown;
+      if (Array.isArray(parsed)) {
+        if (parsed.length === 0) {
+          throw new MACPCapabilityError(
+            'MACP_NO_COMMAND',
+            'gate-spec',
+            `gate spec file '${spec}' contains an empty gates array`,
+          );
+        }
+        return parsed;
+      }
+      if (typeof parsed === 'object' && parsed !== null) {
+        const obj = parsed as Record<string, unknown>;
+        if (Array.isArray(obj['quality_gates'])) {
+          return obj['quality_gates'];
+        }
+        return [parsed];
+      }
+      throw new MACPCapabilityError(
+        'MACP_NO_COMMAND',
+        'gate-spec',
+        `gate spec file '${spec}' parsed to ${typeof parsed} — expected a gates array, a task with quality_gates, or a gate object`,
+      );
+    } catch (exc) {
+      if (exc instanceof MACPCapabilityError) throw exc;
+      // Not JSON — treat each non-empty line as a command gate.
+      const lines = raw
+        .split('\n')
+        .map((l) => l.trim())
+        .filter((l) => l.length > 0);
+      if (lines.length > 0) return lines;
+      throw new MACPCapabilityError(
+        'MACP_NO_COMMAND',
+        'gate-spec',
+        `gate spec file '${spec}' contains no gates`,
+      );
+    }
+  }
+  if (spec.trim().length > 0) return [spec];
+  throw new MACPCapabilityError('MACP_NO_COMMAND', 'gate-spec', 'gate spec is empty');
+}
+
+/** Print a typed not-implemented failure and exit nonzero (RI-N2 fail-closed). */
+function notImplemented(subcommand: string, capability: string, hint: string): void {
+  const err = new MACPCapabilityError(
+    'MACP_NOT_IMPLEMENTED',
+    capability,
+    `${subcommand} is not implemented in @mosaicstack/macp yet (${capability} capability absent) — ${hint}`,
+  );
+  console.error(`[macp] ${subcommand}: ${err.message} [${err.code}]`);
+  process.exitCode = 1;
+}
+
 /**
  * Register macp subcommands on an existing Commander program.
  * This avoids cross-package Commander version mismatches by using the
@@ -24,15 +92,14 @@ export function registerMacpCommand(parent: Command): void {
       'Filter by task type (coding|deploy|research|review|documentation|infrastructure)',
     )
     .action((opts: { status?: string; type?: string }) => {
-      // not yet wired — task persistence layer is not present in @mosaicstack/macp
-      console.log('[macp] tasks list: not yet wired — use macp package programmatically');
+      // unimplemented capability — a failure, never a success (RI-N2)
       if (opts.status) {
         console.log(`  status filter: ${opts.status}`);
       }
       if (opts.type) {
         console.log(`  type filter: ${opts.type}`);
       }
-      process.exitCode = 0;
+      notImplemented('tasks list', 'task-persistence', 'use the macp package programmatically');
     });
 
   // ─── submit ──────────────────────────────────────────────────────────────
@@ -41,12 +108,11 @@ export function registerMacpCommand(parent: Command): void {
     .command('submit <path>')
     .description('Submit a task from a JSON/YAML spec file')
     .action((specPath: string) => {
-      // not yet wired — task submission requires a running MACP server
-      console.log('[macp] submit: not yet wired — use macp package programmatically');
+      // unimplemented capability — a failure, never a success (RI-N2)
       console.log(`  spec path: ${specPath}`);
       console.log('  task id:   (unavailable — no MACP server connected)');
       console.log('  status:    (unavailable — no MACP server connected)');
-      process.exitCode = 0;
+      notImplemented('submit', 'macp-server', 'use the macp package programmatically');
     });
 
   // ─── gate ────────────────────────────────────────────────────────────────
@@ -58,16 +124,58 @@ export function registerMacpCommand(parent: Command): void {
     .option('--cwd <path>', 'Working directory for gate execution', process.cwd())
     .option('--log <path>', 'Path to write gate log output', '/tmp/macp-gate.log')
     .option('--timeout <seconds>', 'Gate timeout in seconds', '60')
-    .action((spec: string, opts: { failOn: string; cwd: string; log: string; timeout: string }) => {
-      // not yet wired — gate execution requires a task context and event sink
-      console.log('[macp] gate: not yet wired — use macp package programmatically');
-      console.log(`  spec:     ${spec}`);
-      console.log(`  fail-on:  ${opts.failOn}`);
-      console.log(`  cwd:      ${opts.cwd}`);
-      console.log(`  log:      ${opts.log}`);
-      console.log(`  timeout:  ${opts.timeout}s`);
-      process.exitCode = 0;
-    });
+    .option(
+      '--simulate',
+      'Simulate gates instead of executing them; results are typed simulated and never satisfy a check',
+    )
+    .action(
+      (
+        spec: string,
+        opts: { failOn: string; cwd: string; log: string; timeout: string; simulate?: boolean },
+      ) => {
+        let gates: unknown[];
+        try {
+          gates = loadGateSpec(spec);
+        } catch (exc) {
+          if (exc instanceof MACPCapabilityError) {
+            console.error(`[macp] gate: ${exc.message} [${exc.code}]`);
+          } else {
+            console.error(`[macp] gate: ${String(exc)}`);
+          }
+          process.exitCode = 1;
+          return;
+        }
+
+        const timeoutSec = Number.parseInt(opts.timeout, 10) || 60;
+        const eventsPath = `${opts.log}.events.ndjson`;
+        const { state, gateResults } = runGates(
+          gates,
+          opts.cwd,
+          opts.log,
+          timeoutSec,
+          eventsPath,
+          'macp-cli-gate',
+          {
+            simulate: opts.simulate,
+          },
+        );
+
+        for (const r of gateResults) {
+          const label = r.command || r.type;
+          const reason = r.reason ? ` — ${r.reason}` : '';
+          console.log(`[macp] gate ${r.status}: ${label}${reason}`);
+        }
+        if (opts.simulate) {
+          console.log(
+            '[macp] SIMULATED run — every result is typed simulated and can never satisfy a gate, dependency, or release check',
+          );
+        }
+
+        // Simulated runs may complete (exit 0) only because the caller
+        // explicitly passed --simulate; the typed state stays 'simulated'.
+        process.exitCode = state === 'passed' || state === 'simulated' ? 0 : 1;
+      },
+    );
 
   // ─── events ──────────────────────────────────────────────────────────────
 
@@ -79,14 +187,16 @@ export function registerMacpCommand(parent: Command): void {
     .option('--file <path>', 'Path to the MACP events NDJSON file')
     .option('--follow', 'Follow the file for new events (like tail -f)')
     .action((opts: { file?: string; follow?: boolean }) => {
-      // not yet wired — event streaming requires a live event source
-      console.log('[macp] events tail: not yet wired — use macp package programmatically');
+      // unimplemented capability — a failure, never a success (RI-N2)
       if (opts.file) {
         console.log(`  file:   ${opts.file}`);
       }
       if (opts.follow) {
         console.log('  mode:   follow');
       }
-      process.exitCode = 0;
+      notImplemented('events tail', 'event-source', 'use the macp package programmatically');
     });
 }
+
+// Re-export so CLI consumers can surface typed capability codes.
+export type { MacpErrorCode };

packages/macp/src/errors.ts

@@ -0,0 +1,35 @@
+/** Typed error code from the closed MACP_ERROR_CODES set. */
+export type MacpErrorCode = (typeof MACP_ERROR_CODES)[number];
+/**
+ * Typed fail-closed capability errors (RI-N2, SDLC-D-035).
+ *
+ * MACP must fail closed when a required capability (executor, reviewer,
+ * command, CI provider, human authority) is absent. These typed codes mirror
+ * the Forge failure vocabulary (FORGE_NO_*) so both packages speak the same
+ * language: an unimplemented capability is a failure, never a stub success.
+ */
+
+/** Closed set of typed MACP capability error codes. */
+export const MACP_ERROR_CODES = [
+  'MACP_NOT_IMPLEMENTED',
+  'MACP_NO_COMMAND',
+  'MACP_NO_REVIEWER',
+  'MACP_NO_CI_PIPELINE',
+  'MACP_NO_PROVIDER',
+  'MACP_AUTHORITY_REQUIRED',
+] as const;
+
+/** Raised when a required capability is missing and execution must fail closed. */
+export class MACPCapabilityError extends Error {
+  /** Typed error code from the closed MACP_ERROR_CODES set. */
+  readonly code: MacpErrorCode;
+  /** The missing capability, e.g. `ci-provider`, `task-persistence`, `command`. */
+  readonly capability: string;
+
+  constructor(code: MacpErrorCode, capability: string, message: string) {
+    super(message);
+    this.name = 'MACPCapabilityError';
+    this.code = code;
+    this.capability = capability;
+  }
+}

packages/macp/src/gate-runner.spec.ts

@@ -0,0 +1,429 @@
+import fs from 'node:fs';
+import os from 'node:os';
+import path from 'node:path';
+import { afterEach, beforeEach, describe, expect, it } from 'vitest';
+
+import { countAIFindings, normalizeGate, runGate, runGates } from './gate-runner.js';
+
+function makeTmpDir(): string {
+  return fs.mkdtempSync(path.join(os.tmpdir(), 'macp-gate-'));
+}
+
+describe('normalizeGate', () => {
+  it('normalizes a string to mechanical gate', () => {
+    expect(normalizeGate('echo test')).toEqual({
+      command: 'echo test',
+      type: 'mechanical',
+      fail_on: 'blocker',
+    });
+  });
+
+  it('normalizes an object gate with defaults', () => {
+    expect(normalizeGate({ command: 'lint' })).toEqual({
+      command: 'lint',
+      type: 'mechanical',
+      fail_on: 'blocker',
+    });
+  });
+
+  it('preserves explicit type and fail_on', () => {
+    expect(normalizeGate({ command: 'review', type: 'ai-review', fail_on: 'any' })).toEqual({
+      command: 'review',
+      type: 'ai-review',
+      fail_on: 'any',
+    });
+  });
+
+  it('handles non-string/non-object input', () => {
+    expect(normalizeGate(42)).toEqual({ command: '', type: 'mechanical', fail_on: 'blocker' });
+    expect(normalizeGate(null)).toEqual({ command: '', type: 'mechanical', fail_on: 'blocker' });
+  });
+});
+
+describe('countAIFindings', () => {
+  it('returns zeros for non-object', () => {
+    expect(countAIFindings(null)).toEqual({ blockers: 0, total: 0 });
+    expect(countAIFindings('string')).toEqual({ blockers: 0, total: 0 });
+    expect(countAIFindings([])).toEqual({ blockers: 0, total: 0 });
+  });
+
+  it('counts from stats block', () => {
+    const output = { stats: { blockers: 2, should_fix: 3, suggestions: 1 } };
+    expect(countAIFindings(output)).toEqual({ blockers: 2, total: 6 });
+  });
+
+  it('counts from findings array when stats has no blockers', () => {
+    const output = {
+      stats: { blockers: 0 },
+      findings: [{ severity: 'blocker' }, { severity: 'warning' }, { severity: 'blocker' }],
+    };
+    expect(countAIFindings(output)).toEqual({ blockers: 2, total: 3 });
+  });
+
+  it('uses stats blockers over findings array when stats has blockers', () => {
+    const output = {
+      stats: { blockers: 5 },
+      findings: [{ severity: 'blocker' }, { severity: 'warning' }],
+    };
+    // stats.blockers = 5, total from stats = 5+0+0 = 5, findings not used for total since stats total is non-zero
+    expect(countAIFindings(output)).toEqual({ blockers: 5, total: 5 });
+  });
+
+  it('counts findings length as total when stats has zero total', () => {
+    const output = {
+      findings: [{ severity: 'warning' }, { severity: 'info' }],
+    };
+    expect(countAIFindings(output)).toEqual({ blockers: 0, total: 2 });
+  });
+});
+
+describe('runGate', () => {
+  let tmp: string;
+  let logPath: string;
+
+  beforeEach(() => {
+    tmp = makeTmpDir();
+    logPath = path.join(tmp, 'gate.log');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmp, { recursive: true, force: true });
+  });
+
+  it('passes mechanical gate on exit 0', () => {
+    const result = runGate('echo hello', tmp, logPath, 30);
+    expect(result.passed).toBe(true);
+    expect(result.exit_code).toBe(0);
+    expect(result.type).toBe('mechanical');
+    expect(result.output).toContain('hello');
+  });
+
+  it('fails mechanical gate on non-zero exit', () => {
+    const result = runGate('exit 1', tmp, logPath, 30);
+    expect(result.passed).toBe(false);
+    expect(result.exit_code).toBe(1);
+  });
+
+  it('ci-pipeline fails closed without a CI provider (no placeholder pass)', () => {
+    const result = runGate({ command: 'anything', type: 'ci-pipeline' }, tmp, logPath, 30);
+    expect(result.passed).toBe(false);
+    expect(result.status).toBe('capability_failure');
+    expect(result.capability_code).toBe('MACP_NO_CI_PIPELINE');
+    expect(result.type).toBe('ci-pipeline');
+    expect(result.output).not.toBe('CI pipeline gate placeholder');
+  });
+
+  it('empty command is a typed capability failure, never a pass', () => {
+    const result = runGate({ command: '' }, tmp, logPath, 30);
+    expect(result.passed).toBe(false);
+    expect(result.status).toBe('capability_failure');
+    expect(result.capability_code).toBe('MACP_NO_COMMAND');
+  });
+
+  it('ai-review gate parses JSON output', () => {
+    const json = JSON.stringify({ stats: { blockers: 0, should_fix: 1 } });
+    const result = runGate({ command: `echo '${json}'`, type: 'ai-review' }, tmp, logPath, 30);
+    expect(result.passed).toBe(true);
+    expect(result.blockers).toBe(0);
+    expect(result.findings).toBe(1);
+  });
+
+  it('ai-review gate fails on blockers', () => {
+    const json = JSON.stringify({ stats: { blockers: 2 } });
+    const result = runGate({ command: `echo '${json}'`, type: 'ai-review' }, tmp, logPath, 30);
+    expect(result.passed).toBe(false);
+    expect(result.blockers).toBe(2);
+  });
+
+  it('ai-review gate with fail_on=any fails on any findings', () => {
+    const json = JSON.stringify({ stats: { blockers: 0, should_fix: 1 } });
+    const result = runGate(
+      { command: `echo '${json}'`, type: 'ai-review', fail_on: 'any' },
+      tmp,
+      logPath,
+      30,
+    );
+    expect(result.passed).toBe(false);
+    expect(result.fail_on).toBe('any');
+  });
+
+  it('ai-review gate fails on invalid JSON output', () => {
+    const result = runGate({ command: 'echo "not json"', type: 'ai-review' }, tmp, logPath, 30);
+    expect(result.passed).toBe(false);
+    expect(result.parse_error).toBeDefined();
+  });
+
+  it('writes to log file', () => {
+    runGate('echo logged', tmp, logPath, 30);
+    const log = fs.readFileSync(logPath, 'utf-8');
+    expect(log).toContain('COMMAND: echo logged');
+    expect(log).toContain('logged');
+    expect(log).toContain('EXIT:');
+  });
+});
+
+describe('runGates', () => {
+  let tmp: string;
+  let logPath: string;
+  let eventsPath: string;
+
+  beforeEach(() => {
+    tmp = makeTmpDir();
+    logPath = path.join(tmp, 'gates.log');
+    eventsPath = path.join(tmp, 'events.ndjson');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmp, { recursive: true, force: true });
+  });
+
+  it('runs multiple gates and returns results', () => {
+    const { allPassed, gateResults } = runGates(
+      ['echo one', 'echo two'],
+      tmp,
+      logPath,
+      30,
+      eventsPath,
+      'task-1',
+    );
+    expect(allPassed).toBe(true);
+    expect(gateResults).toHaveLength(2);
+  });
+
+  it('reports failure when any gate fails', () => {
+    const { allPassed, gateResults } = runGates(
+      ['echo ok', 'exit 1'],
+      tmp,
+      logPath,
+      30,
+      eventsPath,
+      'task-2',
+    );
+    expect(allPassed).toBe(false);
+    expect(gateResults[0]!.passed).toBe(true);
+    expect(gateResults[1]!.passed).toBe(false);
+  });
+
+  it('emits events for each gate', () => {
+    runGates(['echo test'], tmp, logPath, 30, eventsPath, 'task-3');
+    const events = fs
+      .readFileSync(eventsPath, 'utf-8')
+      .trim()
+      .split('\n')
+      .map((l) => JSON.parse(l));
+    expect(events).toHaveLength(2); // started + passed
+    expect(events[0].event_type).toBe('rail.check.started');
+    expect(events[1].event_type).toBe('rail.check.passed');
+  });
+
+  it('does not silently skip gates with empty command — they become capability failures', () => {
+    const { gateResults, allPassed, state } = runGates(
+      [{ command: '', type: 'mechanical' }, 'echo real'],
+      tmp,
+      logPath,
+      30,
+      eventsPath,
+      'task-4',
+    );
+    expect(gateResults).toHaveLength(2);
+    expect(gateResults[0]!.status).toBe('capability_failure');
+    expect(gateResults[1]!.status).toBe('passed');
+    expect(allPassed).toBe(false);
+    expect(state).toBe('capability_failure');
+  });
+
+  it('does not skip ci-pipeline even with empty command — typed capability failure', () => {
+    const { gateResults, allPassed, state } = runGates(
+      [{ command: '', type: 'ci-pipeline' }],
+      tmp,
+      logPath,
+      30,
+      eventsPath,
+      'task-5',
+    );
+    expect(gateResults).toHaveLength(1);
+    expect(gateResults[0]!.passed).toBe(false);
+    expect(gateResults[0]!.status).toBe('capability_failure');
+    expect(allPassed).toBe(false);
+    expect(state).toBe('capability_failure');
+  });
+
+  it('emits failed event with correct message', () => {
+    runGates(['exit 42'], tmp, logPath, 30, eventsPath, 'task-6');
+    const events = fs
+      .readFileSync(eventsPath, 'utf-8')
+      .trim()
+      .split('\n')
+      .map((l) => JSON.parse(l));
+    const failEvent = events.find(
+      (e: Record<string, unknown>) => e.event_type === 'rail.check.failed',
+    );
+    expect(failEvent).toBeDefined();
+    expect(failEvent.message).toContain('Gate failed (');
+  });
+});
+
+/**
+ * RI-N2 / SDLC-D-035 fail-closed controls for the MACP gate runner.
+ *
+ * Invariant under test: `passed: true` occurs ONLY when a gate really executed
+ * and really exited green (`status === 'passed'`). Absent capabilities,
+ * manual sign-offs, and simulated runs are typed distinctly and can never
+ * make the aggregate `passed`.
+ */
+describe('gate-runner fail-closed (RI-N2)', () => {
+  let tmpDir: string;
+  let logPath: string;
+  let eventsPath: string;
+
+  beforeEach(() => {
+    tmpDir = makeTmpDir();
+    logPath = path.join(tmpDir, 'gate.log');
+    eventsPath = path.join(tmpDir, 'events.ndjson');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  function run(gates: unknown[], options?: { simulate?: boolean }) {
+    return runGates(gates, tmpDir, logPath, 10, eventsPath, 'spec-task', options);
+  }
+
+  // ─── positive controls ───────────────────────────────────────────────────
+
+  it('a really-executed green command gate still passes', () => {
+    const result = run([{ command: 'exit 0', type: 'mechanical' }]);
+    expect(result.gateResults[0]!.status).toBe('passed');
+    expect(result.gateResults[0]!.passed).toBe(true);
+    expect(result.allPassed).toBe(true);
+    expect(result.state).toBe('passed');
+  });
+
+  it('explicit simulate completes and types every result simulated', () => {
+    const result = run([{ command: 'exit 0', type: 'mechanical' }, 'echo hello'], {
+      simulate: true,
+    });
+    expect(result.gateResults).toHaveLength(2);
+    for (const gate of result.gateResults) {
+      expect(gate.status).toBe('simulated');
+      expect(gate.passed).toBe(false);
+    }
+    expect(result.state).toBe('simulated');
+  });
+
+  it('a really-executed red command gate fails with typed status failed', () => {
+    const result = run([{ command: 'exit 3', type: 'mechanical' }]);
+    expect(result.gateResults[0]!.status).toBe('failed');
+    expect(result.gateResults[0]!.passed).toBe(false);
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('failed');
+  });
+
+  // ─── negative controls — each asserts typed status AND aggregate not passed ──
+
+  it('an empty-command gate is a capability_failure, not skipped and not passed', () => {
+    const result = run([{ command: '', type: 'mechanical' }]);
+    // runGates must not silently skip it — it produces a typed result
+    expect(result.gateResults).toHaveLength(1);
+    const gate = result.gateResults[0]!;
+    expect(gate.status).toBe('capability_failure');
+    expect(gate.capability_code).toBe('MACP_NO_COMMAND');
+    expect(gate.passed).toBe(false);
+    // aggregate is not passed
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('capability_failure');
+    expect(result.state).not.toBe('passed');
+  });
+
+  it('a commandless ai-review gate is a typed MACP_NO_REVIEWER capability_failure', () => {
+    const result = run([{ command: '', type: 'ai-review' }]);
+    expect(result.gateResults[0]!.status).toBe('capability_failure');
+    expect(result.gateResults[0]!.capability_code).toBe('MACP_NO_REVIEWER');
+    expect(result.allPassed).toBe(false);
+    expect(result.state).not.toBe('passed');
+  });
+
+  it('a ci-pipeline gate without a provider implementation is a capability_failure, never a placeholder pass', () => {
+    const result = run([{ command: '', type: 'ci-pipeline' }]);
+    const gate = result.gateResults[0]!;
+    expect(gate.status).toBe('capability_failure');
+    expect(gate.capability_code).toBe('MACP_NO_CI_PIPELINE');
+    expect(gate.passed).toBe(false);
+    // the old false-success placeholder must be gone
+    expect(gate.output).not.toBe('CI pipeline gate placeholder');
+    expect(result.allPassed).toBe(false);
+    expect(result.state).not.toBe('passed');
+  });
+
+  it('a ci-pipeline gate fails closed even alongside an otherwise green run', () => {
+    const result = run(['exit 0', { type: 'ci-pipeline', command: 'fake-ci' }]);
+    expect(result.gateResults[1]!.status).toBe('capability_failure');
+    expect(result.gateResults[0]!.status).toBe('passed');
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('capability_failure');
+  });
+
+  it('a manual gate with no automation enters typed waiting — neither pass nor fail', () => {
+    const result = run([{ type: 'manual' }]);
+    const gate = result.gateResults[0]!;
+    expect(gate.status).toBe('waiting');
+    expect(gate.passed).toBe(false);
+    expect(gate.exit_code).toBe(0);
+    // aggregate is not passed while any gate is waiting
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('waiting');
+    expect(result.state).not.toBe('passed');
+  });
+
+  it('a simulated result can never make the aggregate passed', () => {
+    const result = run(['exit 0', 'exit 0'], { simulate: true });
+    expect(result.gateResults.every((g) => g.status === 'simulated')).toBe(true);
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('simulated');
+    expect(result.state).not.toBe('passed');
+  });
+
+  it('waiting dominates an otherwise green aggregate', () => {
+    const result = run(['exit 0', { type: 'manual' }]);
+    expect(result.allPassed).toBe(false);
+    expect(result.state).toBe('waiting');
+  });
+});
+
+describe('runGate fail-closed (RI-N2)', () => {
+  let tmpDir: string;
+  let logPath: string;
+
+  beforeEach(() => {
+    tmpDir = makeTmpDir();
+    logPath = path.join(tmpDir, 'gate.log');
+  });
+
+  afterEach(() => {
+    fs.rmSync(tmpDir, { recursive: true, force: true });
+  });
+
+  it('simulate: true returns a typed simulated result without executing', () => {
+    const result = runGate('this-command-does-not-exist-xyz', tmpDir, logPath, 10, {
+      simulate: true,
+    });
+    expect(result.status).toBe('simulated');
+    expect(result.passed).toBe(false);
+    expect(result.exit_code).toBe(0);
+  });
+
+  it('normal mode executes for real and types a green gate passed', () => {
+    const result = runGate('echo ok', tmpDir, logPath, 10);
+    expect(result.status).toBe('passed');
+    expect(result.passed).toBe(true);
+    expect(result.output).toContain('ok');
+  });
+
+  it('a bare string gate normalizes to mechanical and executes', () => {
+    const result = runGate('exit 7', tmpDir, logPath, 10);
+    expect(result.type).toBe('mechanical');
+    expect(result.status).toBe('failed');
+    expect(result.passed).toBe(false);
+  });
+});

packages/macp/src/gate-runner.ts

@@ -4,7 +4,20 @@ import { dirname } from 'node:path';
 
 import { emitEvent } from './event-emitter.js';
 import { nowISO } from './event-emitter.js';
-import type { GateResult } from './types.js';
+import type { GateResult, GateStatus, RunGatesResult } from './types.js';
+
+/** Typed reason stamped on every simulated gate result. */
+export const SIMULATED_GATE_REASON =
+  'simulated execution (explicit simulate opt-in): gate was not evaluated by a real implementation';
+
+/** Options for gate execution (RI-N2 fail-closed / explicit simulation). */
+export interface RunGateOptions {
+  /**
+   * Explicit caller opt-in to simulation. Simulated gates are NOT executed;
+   * every result is typed `simulated` and never satisfies anything.
+   */
+  simulate?: boolean;
+}
 
 export interface NormalizedGate {
   command: string;
@@ -103,36 +116,91 @@ export function countAIFindings(parsedOutput: unknown): { blockers: number; tota
   return { blockers, total };
 }
 
+function simulatedResult(gateEntry: NormalizedGate): GateResult {
+  return {
+    command: gateEntry.command,
+    exit_code: 0,
+    type: gateEntry.type,
+    output: SIMULATED_GATE_REASON,
+    timed_out: false,
+    passed: false,
+    status: 'simulated',
+    reason: SIMULATED_GATE_REASON,
+  };
+}
+
+function capabilityFailureResult(
+  gateEntry: NormalizedGate,
+  code: GateResult['capability_code'],
+  reason: string,
+): GateResult {
+  return {
+    command: gateEntry.command,
+    exit_code: 1,
+    type: gateEntry.type,
+    output: '',
+    timed_out: false,
+    passed: false,
+    status: 'capability_failure',
+    capability_code: code,
+    reason,
+  };
+}
+
+function waitingResult(gateEntry: NormalizedGate, reason: string): GateResult {
+  return {
+    command: gateEntry.command,
+    exit_code: 0,
+    type: gateEntry.type,
+    output: '',
+    timed_out: false,
+    passed: false,
+    status: 'waiting',
+    capability_code: 'MACP_AUTHORITY_REQUIRED',
+    reason,
+  };
+}
+
 export function runGate(
   gate: unknown,
   cwd: string,
   logPath: string,
   timeoutSec: number,
+  options: RunGateOptions = {},
 ): GateResult {
   const gateEntry = normalizeGate(gate);
   const gateType = gateEntry.type;
   const command = gateEntry.command;
 
+  // Explicit simulation only: never executes, typed simulated, never satisfying.
+  if (options.simulate) {
+    return simulatedResult(gateEntry);
+  }
+
+  // Fail closed: no CI provider implementation exists in @mosaicstack/macp,
+  // so a ci-pipeline gate is an absent capability — never a placeholder pass.
   if (gateType === 'ci-pipeline') {
-    return {
-      command,
-      exit_code: 0,
-      type: gateType,
-      output: 'CI pipeline gate placeholder',
-      timed_out: false,
-      passed: true,
-    };
+    return capabilityFailureResult(
+      gateEntry,
+      'MACP_NO_CI_PIPELINE',
+      `ci-pipeline gate '${gateEntry.command || gateType}' has no CI provider implementation wired — refusing placeholder pass`,
+    );
   }
 
   if (!command) {
-    return {
-      command: '',
-      exit_code: 0,
-      type: gateType,
-      output: '',
-      timed_out: false,
-      passed: true,
-    };
+    // A manual gate with no automation waits for human sign-off: not pass, not fail.
+    if (gateType === 'manual') {
+      return waitingResult(
+        gateEntry,
+        `manual gate has no automation — waiting for human sign-off (type: ${gateType})`,
+      );
+    }
+    // Any other commandless gate is an absent capability — never a vacuous pass.
+    return capabilityFailureResult(
+      gateEntry,
+      gateType === 'ai-review' ? 'MACP_NO_REVIEWER' : 'MACP_NO_COMMAND',
+      `gate of type '${gateType}' has no command to execute — refusing empty-command pass`,
+    );
   }
 
   const { exitCode, output, timedOut } = runShell(command, cwd, logPath, timeoutSec);
@@ -143,10 +211,12 @@ export function runGate(
     output,
     timed_out: timedOut,
     passed: false,
+    status: 'failed',
   };
 
   if (gateType !== 'ai-review') {
     result.passed = exitCode === 0;
+    result.status = result.passed ? 'passed' : 'failed';
     return result;
   }
 
@@ -170,6 +240,7 @@ export function runGate(
   } else {
     result.passed = exitCode === 0 && blockers === 0 && !timedOut && parseError === undefined;
   }
+  result.status = result.passed ? 'passed' : 'failed';
 
   result.fail_on = failOn;
   result.blockers = blockers;
@@ -191,16 +262,19 @@ export function runGates(
   timeoutSec: number,
   eventsPath: string,
   taskId: string,
-): { allPassed: boolean; gateResults: GateResult[] } {
-  let allPassed = true;
+  options: RunGateOptions = {},
+): RunGatesResult {
   const gateResults: GateResult[] = [];
+  let hasCapabilityFailure = false;
+  let hasSimulated = false;
+  let hasFailed = false;
+  let hasWaiting = false;
 
   for (const gate of gates) {
     const gateEntry = normalizeGate(gate);
     const gateCmd = gateEntry.command;
-    if (!gateCmd && gateEntry.type !== 'ci-pipeline') continue;
-
     const label = gateCmd || gateEntry.type;
+    // NOTE: no silent skip — every gate produces a typed result (RI-N2).
     emitEvent(
       eventsPath,
       'rail.check.started',
@@ -209,10 +283,10 @@ export function runGates(
       'quality-gate',
       `Running gate: ${label}`,
     );
-    const result = runGate(gate, cwd, logPath, timeoutSec);
+    const result = runGate(gate, cwd, logPath, timeoutSec, options);
     gateResults.push(result);
 
-    if (result.passed) {
+    if (result.status === 'passed') {
       emitEvent(
         eventsPath,
         'rail.check.passed',
@@ -224,7 +298,46 @@ export function runGates(
       continue;
     }
 
-    allPassed = false;
+    if (result.status === 'waiting') {
+      hasWaiting = true;
+      emitEvent(
+        eventsPath,
+        'rail.check.waiting',
+        taskId,
+        'gated',
+        'quality-gate',
+        `Gate waiting: ${label} — ${result.reason ?? 'manual gate awaits sign-off'}`,
+      );
+      continue;
+    }
+
+    if (result.status === 'simulated') {
+      hasSimulated = true;
+      emitEvent(
+        eventsPath,
+        'rail.check.simulated',
+        taskId,
+        'gated',
+        'quality-gate',
+        `Gate simulated (non-satisfying): ${label}`,
+      );
+      continue;
+    }
+
+    if (result.status === 'capability_failure') {
+      hasCapabilityFailure = true;
+      emitEvent(
+        eventsPath,
+        'rail.check.failed',
+        taskId,
+        'gated',
+        'quality-gate',
+        `Gate capability failure (${result.capability_code ?? 'MACP_NO_PROVIDER'}): ${label} — ${result.reason ?? 'required capability is absent'}`,
+      );
+      continue;
+    }
+
+    hasFailed = true;
     let message: string;
     if (result.timed_out) {
       message = `Gate timed out after ${timeoutSec}s: ${label}`;
@@ -236,5 +349,15 @@ export function runGates(
     emitEvent(eventsPath, 'rail.check.failed', taskId, 'gated', 'quality-gate', message);
   }
 
-  return { allPassed, gateResults };
+  const state: GateStatus = hasCapabilityFailure
+    ? 'capability_failure'
+    : hasSimulated
+      ? 'simulated'
+      : hasFailed
+        ? 'failed'
+        : hasWaiting
+          ? 'waiting'
+          : 'passed';
+
+  return { allPassed: state === 'passed', gateResults, state };
 }

packages/macp/src/index.ts

@@ -6,11 +6,13 @@ export type {
   DependsOnPolicy,
   GateType,
   GateFailOn,
+  GateStatus,
   GateEntry,
   Task,
   EventType,
   MACPEvent,
   GateResult,
+  RunGatesResult,
   TaskResult,
   ProviderMeta,
   ProviderRegistry,
@@ -18,6 +20,11 @@ export type {
 
 export { CredentialError } from './types.js';
 
+// Typed fail-closed capability errors (RI-N2, SDLC-D-035)
+export { MACP_ERROR_CODES, MACPCapabilityError } from './errors.js';
+
+export type { MacpErrorCode } from './errors.js';
+
 // Credential resolver
 export {
   DEFAULT_CREDENTIALS_DIR,
@@ -35,9 +42,16 @@ export {
 export type { ResolveCredentialsOptions } from './credential-resolver.js';
 
 // Gate runner
-export { normalizeGate, runShell, countAIFindings, runGate, runGates } from './gate-runner.js';
+export {
+  normalizeGate,
+  runShell,
+  countAIFindings,
+  runGate,
+  runGates,
+  SIMULATED_GATE_REASON,
+} from './gate-runner.js';
 
-export type { NormalizedGate } from './gate-runner.js';
+export type { NormalizedGate, RunGateOptions } from './gate-runner.js';
 
 // Risk-floor (agent reflection loop — diff review classifier)
 export { evaluateRiskFloor, DEFAULT_RISK_THRESHOLD } from './risk-floor.js';

packages/macp/src/types.ts

@@ -1,3 +1,5 @@
+import type { MacpErrorCode } from './errors.js';
+
 /** Task status values. */
 export type TaskStatus = 'pending' | 'running' | 'gated' | 'completed' | 'failed' | 'escalated';
 
@@ -17,7 +19,17 @@ export type DispatchMode = 'yolo' | 'acp' | 'exec';
 export type DependsOnPolicy = 'all' | 'any' | 'all_terminal';
 
 /** Quality gate type. */
-export type GateType = 'mechanical' | 'ai-review' | 'ci-pipeline';
+export type GateType = 'mechanical' | 'ai-review' | 'ci-pipeline' | 'manual';
+
+/**
+ * Typed execution state of a gate — closed set (RI-N2, SDLC-D-035).
+ *
+ * Only `passed` means "really executed and green". `simulated` is produced
+ * exclusively under an explicit simulate opt-in and never satisfies anything.
+ * `capability_failure` means a required executor/provider/command was absent.
+ * `waiting` means a manual gate awaits human sign-off (neither pass nor fail).
+ */
+export type GateStatus = 'passed' | 'failed' | 'simulated' | 'waiting' | 'capability_failure';
 
 /** Gate fail_on mode. */
 export type GateFailOn = 'blocker' | 'any';
@@ -67,7 +79,9 @@ export type EventType =
   | 'task.retry.scheduled'
   | 'rail.check.started'
   | 'rail.check.passed'
-  | 'rail.check.failed';
+  | 'rail.check.failed'
+  | 'rail.check.waiting'
+  | 'rail.check.simulated';
 
 /** Structured event record. */
 export interface MACPEvent {
@@ -88,7 +102,14 @@ export interface GateResult {
   type: string;
   output: string;
   timed_out: boolean;
+  /** Back-compat boolean view — true ONLY when `status === 'passed'`. */
   passed: boolean;
+  /** Typed discriminator — the authoritative gate outcome (RI-N2). */
+  status: GateStatus;
+  /** Typed capability error code, set when `status === 'capability_failure'`. */
+  capability_code?: MacpErrorCode;
+  /** Why a non-executed state (simulated/waiting/capability_failure) was reached. */
+  reason?: string;
   fail_on?: string;
   blockers?: number;
   findings?: number;
@@ -96,6 +117,22 @@ export interface GateResult {
   parse_error?: string;
 }
 
+/**
+ * Aggregate outcome of `runGates` (RI-N2).
+ *
+ * `state` is the typed aggregate: it is `passed` only when every gate really
+ * executed green. A `simulated` result makes the aggregate `simulated` (never
+ * `passed`); a `waiting` manual gate keeps the aggregate `waiting`; a missing
+ * capability makes it `capability_failure`. `allPassed` is exactly
+ * `state === 'passed'`, so a simulated or waiting result can never satisfy a
+ * dependency, acceptance criterion, gate, merge, or release check.
+ */
+export interface RunGatesResult {
+  allPassed: boolean;
+  gateResults: GateResult[];
+  state: GateStatus;
+}
+
 /** Result from a completed task. */
 export interface TaskResult {
   task_id: string;

packages/mosaic/__tests__/platform/detect.test.ts

@@ -0,0 +1,74 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+
+// homedir/platform are read at call time, so they can be stubbed per case.
+vi.mock('node:os', async (importOriginal) => {
+  const actual = await importOriginal<typeof import('node:os')>();
+  return {
+    ...actual,
+    homedir: () => '/home/tester',
+    platform: () => mockPlatform,
+  };
+});
+
+let mockPlatform: NodeJS.Platform = 'linux';
+
+const { getShellProfilePath, detectShell } = await import('../../src/platform/detect.js');
+
+describe('getShellProfilePath', () => {
+  const originalShell = process.env['SHELL'];
+  const originalZdotdir = process.env['ZDOTDIR'];
+
+  beforeEach(() => {
+    mockPlatform = 'linux';
+    delete process.env['ZDOTDIR'];
+  });
+
+  afterEach(() => {
+    if (originalShell === undefined) delete process.env['SHELL'];
+    else process.env['SHELL'] = originalShell;
+    if (originalZdotdir === undefined) delete process.env['ZDOTDIR'];
+    else process.env['ZDOTDIR'] = originalZdotdir;
+  });
+
+  // The regression this guards: setupPath() in stages/finalize.ts appends the
+  // PATH export to whatever this returns. A line written to ~/.bashrc is
+  // unreachable to `bash -lc`, systemd units and agent seats, because Debian's
+  // default .bashrc returns early for non-interactive shells — so an install
+  // reported success and left `mosaic: command not found`. Same for .zshrc,
+  // which zsh only reads for interactive shells.
+  it('never targets an interactive-only rc file', () => {
+    for (const shell of ['/bin/bash', '/usr/bin/zsh']) {
+      process.env['SHELL'] = shell;
+      const profile = getShellProfilePath();
+      expect(profile).not.toMatch(/\.bashrc$/);
+      expect(profile).not.toMatch(/\.zshrc$/);
+    }
+  });
+
+  it('uses ~/.profile for bash', () => {
+    process.env['SHELL'] = '/bin/bash';
+    expect(getShellProfilePath()).toBe('/home/tester/.profile');
+  });
+
+  it('uses ~/.zshenv for zsh', () => {
+    process.env['SHELL'] = '/usr/bin/zsh';
+    expect(getShellProfilePath()).toBe('/home/tester/.zshenv');
+  });
+
+  it('honours ZDOTDIR for zsh', () => {
+    process.env['SHELL'] = '/usr/bin/zsh';
+    process.env['ZDOTDIR'] = '/custom/zdot';
+    expect(getShellProfilePath()).toBe('/custom/zdot/.zshenv');
+  });
+
+  it('falls back to ~/.profile for an unknown shell', () => {
+    process.env['SHELL'] = '/bin/somethingelse';
+    expect(detectShell()).toBe('unknown');
+    expect(getShellProfilePath()).toBe('/home/tester/.profile');
+  });
+
+  it('still routes fish to its own config', () => {
+    process.env['SHELL'] = '/usr/bin/fish';
+    expect(getShellProfilePath()).toBe('/home/tester/.config/fish/config.fish');
+  });
+});

packages/mosaic/framework/defaults/AGENTS.md

@@ -43,6 +43,8 @@ overwritten on upgrade. (Layer model: `constitution/LAYER-MODEL.md`.)
 | Secrets / vault usage                              | `guides/VAULT-SECRETS.md`          |
 | Tool/credential reference (service CLIs, wrappers) | `guides/TOOLS-REFERENCE.md`        |
 | Memory protocol (OpenBrain capture/recall)         | `guides/MEMORY.md`                 |
+| Seat identity, git credentials, token slots        | `guides/SEAT-IDENTITY.md`          |
+| Reaching another agent (fleet comms)               | `guides/FLEET-COMMS.md`            |
 
 ## Subagent Model Selection (Cost — Hard Rule)
 
@@ -82,6 +84,7 @@ is re-seeded a genuinely missing core file is a stop-and-report condition — no
 
 Confirm: required + situational tests passed (primary gate); aligned to `docs/PRD.md`; acceptance
 criteria mapped to evidence; independent code review passed (if code changed); required docs updated;
-scratchpad updated. For PR-workflow delivery: merged PR number + merge commit on `main`, terminal-green
+scratchpad updated. For PR-workflow delivery: merged PR number + merge commit on the integration
+trunk (the project's declared trunk, default `main` — see `CONSTITUTION.md` Hard Gates), terminal-green
 CI, linked issue closed (or `docs/TASKS.md` equivalent). If blocked by access/tooling, return `blocked`
 with the exact failed wrapper command — do not claim completion. Full checklist: `guides/E2E-DELIVERY.md`.

packages/mosaic/framework/defaults/CONSTITUTION.md

@@ -21,11 +21,25 @@ guard"), the runtime adapter binds it to a concrete tool and states whether abse
 
 ## Hard Gates
 
+The **integration trunk** is the branch a project declares in its `.mosaic/repo.json` under the
+key `integration_trunk`; `release_branch` names the release target when one exists (`null` for
+single-branch projects). Absent a declaration, the trunk is `main`. The declaration is policy
+data, never shell text: values must be valid local branch names under `git check-ref-format
+--branch` semantics — no remote refs, no revision expressions, no option-like values (leading `-`),
+no path traversal or control characters. A declaration file that fails to parse, an unknown or
+misspelled key, or an invalid value is a hard stop (`blocked`) — never a silent fallback to `main`.
+Prose that mentions branch names designates nothing; only the declaration file does. A project
+declares exactly ONE trunk. **Changing an existing declaration is operator-owned:** a trunk
+redeclaration redirects merge target and branch-protection target at once, so it requires an
+explicit operator action above ordinary PR review. The designation relaxes nothing:
+reviewed-PR-only delivery, squash merge, independent review, queue guards, and terminal-green CI
+bind to the declared trunk exactly as they bind to `main`.
+
 1. Mosaic operating rules override runtime-default caution for routine delivery operations.
 2. Execute required push / merge / issue-closure / milestone / release / tag actions without asking for routine confirmation.
 3. Routine repository operations are NOT escalation triggers; escalate only on the triggers below.
 4. For source-code delivery, completion is forbidden at the PR-open stage.
-5. Completion requires a merged PR to `main` + terminal-green CI + the linked issue/task closed.
+5. Completion requires a merged PR to the integration trunk + terminal-green CI + the linked issue/task closed.
 6. Before any push or merge, run the CI queue guard.
 7. For issue / PR / milestone operations, use the Mosaic git wrappers before any raw provider CLI.
 8. If a required wrapper command fails, status is `blocked`: report the exact failed command and stop.
@@ -35,7 +49,7 @@ guard"), the runtime adapter binds it to a concrete tool and states whether abse
 12. The intake procedure is not conditional on perceived complexity; a "simple" task carries the same requirements as a multi-file feature.
 13. **Merge authority (coordinated work):** when a coordinator/orchestrator session is active for the work, the post-review merge go-ahead is the coordinator's to give — once the required review gates pass, merge on the coordinator's confirmation; do not wait on the human owner personally. Solo (uncoordinated) delivery keeps the default: merge per gates 2 and 9. A "No self-merge" note on a PR means no UNREVIEWED self-merge — it does not suspend coordinator-authorized merges.
 14. Never hardcode secrets; never emit credential values in any output (not even partially, not "to confirm").
-15. Trunk-based git only: branch from `main`, merge via a reviewed PR (squash), never push directly to `main`.
+15. Trunk-based git only: branch from the integration trunk, merge via a reviewed PR (squash), never push directly to the trunk.
 16. If you modify source code, an independent review (author ≠ reviewer) must pass before completion.
 
 ## Integrity (quality gates are never bypassed)

packages/mosaic/framework/defaults/README.md

@@ -133,7 +133,7 @@ You can still launch runtimes directly (`claude`, `codex`, etc.) — thin runtim
 │   ├── opencode/          ← AGENTS.md, RUNTIME.md
 │   ├── pi/                ← RUNTIME.md, mosaic-extension.ts, goal-extension.ts
 │   └── mcp/               ← MCP server configs
-├── skills/                ← Universal skills (synced from mosaic/agent-skills)
+├── skills/                ← Universal skills (shipped with the framework package)
 ├── skills-local/          ← Local cross-runtime skills
 ├── memory/                ← Persistent agent memory (preserved across upgrades)
 └── templates/             ← SOUL.md template, project templates
@@ -200,11 +200,11 @@ The installer rejects unrecognized flags or positional arguments before making c
 
 ## Universal Skills
 
-The installer syncs skills from `mosaic/agent-skills` into `~/.config/mosaic/skills/`. Install, wizard finalization, and `mosaic update` automatically reconcile every canonical skill into Claude Code's `~/.claude/skills/` directory.
+Canonical skills ship inside the framework package itself; the installer installs them into `~/.config/mosaic/skills/` together with the rest of the framework (there is no separate skills repository). Install, wizard finalization, and `mosaic update` automatically link every canonical skill into Claude Code's `~/.claude/skills/` directory.
 
 ```bash
-mosaic sync                                                   # Full canonical catalog sync
-~/.config/mosaic/tools/_scripts/mosaic-sync-skills --link-only # Re-link only
+mosaic sync                                                   # Relink the full canonical catalog
+~/.config/mosaic/tools/_scripts/mosaic-sync-skills --link-only # Re-link only (same as default)
 mosaic skill list                                              # Show registered, missing, dangling, and foreign entries
 mosaic skill register <name>                                   # Register or repair one canonical Claude link
 mosaic skill unregister <name>                                 # Remove one Mosaic-owned Claude link

packages/mosaic/framework/fleet/README.md

@@ -12,6 +12,33 @@ The default tmux socket is `mosaic-fleet` so fleet commands do not touch the
 default tmux server. The roster is the desired-state authority; generated environment files are
 rebuildable projections, never a second source of configuration.
 
+## Brain-home split (fleet state vs framework templates)
+
+When a mosaic-brain clone is present, fleet **state** resolves from the brain
+home while framework templates and dispatch state stay in the config home
+(three-tree model, canon `docs/STRUCTURE-CANON.md` §2):
+
+| Path                                                                            | Without brain (legacy)                | With brain                     |
+| ------------------------------------------------------------------------------- | ------------------------------------- | ------------------------------ |
+| `fleet/agents/<seat>.env.*`                                                     | `~/.config/mosaic/fleet/agents/`      | `~/.mosaic/fleet/agents/`      |
+| `fleet/roles.local/` (overrides)                                                | `~/.config/mosaic/fleet/roles.local/` | `~/.mosaic/fleet/roles.local/` |
+| `fleet/profiles/` (working copies)                                              | `~/.config/mosaic/fleet/profiles/`    | `~/.mosaic/fleet/profiles/`    |
+| `fleet/roster.yaml`, `fleet/roles/` (baseline), `fleet/run/`, `fleet/services/` | `~/.config/mosaic/fleet/…`            | unchanged (config home)        |
+
+Activation (`packages/mosaic/src/fleet/brain-home.ts`, mirrored in
+`tools/fleet/start-agent-session.sh`):
+
+1. `MOSAIC_BRAIN_HOME` env var — explicit, always wins.
+2. Canonical `~/.mosaic` — adopted only when `MOSAIC_HOME` is the default
+   `~/.config/mosaic` AND `~/.mosaic/fleet/agents` exists. Custom
+   `--mosaic-home` values (tests, sandboxes, canaries) never adopt, keeping
+   them hermetic.
+3. Otherwise the config home (legacy single-tree behavior).
+
+Seat env dirs under a brain are subject to the same privacy boundary (0700
+dirs, 0600 files); `.env.generated` files are structure-valuable and tracked
+in the brain repo, hand-maintained `.env`/`.env.local` stay ignored and private.
+
 ## Examples
 
 - `examples/minimal.yaml` starts one local canary slot.

packages/mosaic/framework/fleet/roles/board.md

@@ -35,4 +35,4 @@ A standing panel of senior voices, each arguing from a fixed vantage. The board
 deliberately slow and adversarial — its value is catching the expensive mistake
 before a single agent-hour is spent on it.
 
-> Doctrine: `docs/fleet/north-star.md` ('board' role = forge BOD; role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` ('board' role = forge BOD; role library).

packages/mosaic/framework/fleet/roles/code.md

@@ -33,4 +33,4 @@ The code role writes the change and opens the PR; it never touches the merge pat
 The focused builder. It takes one well-scoped card, drives it to green, opens a
 clean PR, and hands off — never reaching past the card it was given.
 
-> Doctrine: `docs/fleet/north-star.md` (role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library).

packages/mosaic/framework/fleet/roles/decomposition.md

@@ -35,4 +35,4 @@ The work-breakdown specialist. It takes a phased plan and a DAG and emits a clea
 linked set of single-PR cards on the Mosaic backlog — then steps back and lets the
 executors run.
 
-> Doctrine: `docs/fleet/north-star.md` (role library); spend accounting is a process mandate.
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library); spend accounting is a process mandate.

packages/mosaic/framework/fleet/roles/documentation.md

@@ -36,4 +36,4 @@ The scribe of record. It makes sure the docs and the north star's projections
 describe the system as it actually is, and it never lets two writers fight over one
 TASKS file.
 
-> Doctrine: `docs/fleet/north-star.md` (role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library).

packages/mosaic/framework/fleet/roles/enhancer.md

@@ -38,4 +38,4 @@ The orchestrator alone optimizes for _this_ delivery; the enhancer optimizes for
 delivery — self-healing the fleet's tools, skills, and harnesses, and routing real defects
 upstream. Together they are the irreducible core; every other role is added on demand.
 
-> Doctrine: `docs/fleet/north-star.md` (two-agent floor + role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (two-agent floor + role library).

packages/mosaic/framework/fleet/roles/merge-gate.md

@@ -46,4 +46,4 @@ The single, accountable gatekeeper. It waits for green CI (`pr-ci-wait.sh`),
 respects the pause switch, merges only through `pr-merge.sh`, and records every
 decision — so the fleet has exactly one trustworthy door to production.
 
-> Doctrine: `docs/fleet/north-star.md` (role library); merge path: `pr-merge.sh` + `pr-ci-wait.sh`; forbidden paths: `pr-merge.sh` guard.
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library); merge path: `pr-merge.sh` + `pr-ci-wait.sh`; forbidden paths: `pr-merge.sh` guard.

packages/mosaic/framework/fleet/roles/operator.md

@@ -35,4 +35,4 @@ The on-call dispatcher. It makes sure every escalation is seen and re-seen until
 handled, and it holds the one switch that can stop the fleet when something is
 wrong.
 
-> Doctrine: `docs/fleet/north-star.md` (role library); pause switch: `fleet/run/PAUSED`.
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library); pause switch: `fleet/run/PAUSED`.

packages/mosaic/framework/fleet/roles/orchestrator.md

@@ -43,4 +43,4 @@ A lean, decisive coordinator. It thinks in readiness and throughput, dispatches
 next ready card the instant a dependency clears, and never lets an idle agent sit
 while ready work exists — keeping its own context minimal so the loop never slows.
 
-> Doctrine: `docs/fleet/north-star.md` (two-agent floor + role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (two-agent floor + role library).

packages/mosaic/framework/fleet/roles/planner.md

@@ -38,4 +38,4 @@ The architect of the mission's shape. It thinks in phases and dependencies, hand
 a clean DAG to decomposition, and reports its plan back to the orchestrator that
 dispatched it.
 
-> Doctrine: `docs/fleet/north-star.md` (two-agent floor + role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (two-agent floor + role library).

packages/mosaic/framework/fleet/roles/rebase.md

@@ -34,4 +34,4 @@ The rebase role keeps PR branches fresh; it never approves or merges.
 The janitor of the merge queue. It quietly keeps branches current and re-runnable,
 and knows when a conflict is beyond a mechanical rebase and must be escalated.
 
-> Doctrine: `docs/fleet/north-star.md` (role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library).

packages/mosaic/framework/fleet/roles/review.md

@@ -35,4 +35,4 @@ The careful reader. It assumes nothing, checks the change against its card and i
 tests, and is willing to say "not yet" — its value is catching the wrong change
 before it reaches the merge-gate.
 
-> Doctrine: `docs/fleet/north-star.md` (role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library).

packages/mosaic/framework/fleet/roles/security-review.md

@@ -36,4 +36,4 @@ tree or the merge path.
 The adversary on your side. It reads every diff asking "how does this get exploited
 or leak?" — the second, security-focused pair of eyes before the merge-gate.
 
-> Doctrine: `docs/fleet/north-star.md` (role library); forbidden paths: `pr-merge.sh` guard.
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library); forbidden paths: `pr-merge.sh` guard.

packages/mosaic/framework/fleet/roles/session-review.md

@@ -34,4 +34,4 @@ The retrospective analyst. It reads completed sessions and distills them into cl
 actionable signals — the raw material the enhancer uses to make the fleet better
 next time.
 
-> Doctrine: `docs/fleet/north-star.md` (role library); consumed by the enhancer role.
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library); consumed by the enhancer role.

packages/mosaic/framework/fleet/roles/site-tester.md

@@ -34,4 +34,4 @@ merge path.
 The skeptic who insists on running it. It trusts observed behavior over claimed
 behavior, and turns "should work" into "verified works" — or a concrete bug report.
 
-> Doctrine: `docs/fleet/north-star.md` (role library).
+> Doctrine: `docs/fleet/FLEET-DOCTRINE.md` (role library).

packages/mosaic/framework/guides/BOOTSTRAP.md

@@ -12,7 +12,7 @@ This guide covers how to bootstrap a project so AI agents (Claude, Codex, etc.)
 4. Issue tracking is consistent across projects
 5. Documentation standards and API contracts are enforced from day one
 6. PRD requirements are established before coding begins
-7. Branching/merging is consistent: `branch -> main` via PR with squash-only merges
+7. Branching/merging is consistent: branch -> integration trunk (default `main`) via PR with squash-only merges
 8. Steered-autonomy execution is enabled so agents can run end-to-end with escalation-only human intervention
 
 ## Agent Host Prerequisites
@@ -206,7 +206,7 @@ Every runtime context file should contain:
 6. **Issue tracking** — Issue and commit conventions
 7. **Code review** — Required review process
 8. **Runtime notes** — Runtime-specific behavior references
-9. **Branch and merge policy** — Trunk workflow (`branch -> main` via PR, squash-only)
+9. **Branch and merge policy** — Trunk workflow (branch -> integration trunk via PR, squash-only)
 10. **Autonomy and escalation policy** — Agent owns coding/review/PR/release/deploy lifecycle
 
 ---
@@ -288,15 +288,17 @@ Reserve `0.1.0` for the MVP release milestone.
 
 ---
 
-## Step 5b: Configure Main Branch Protection (Hard Rule)
+## Step 5b: Configure Trunk Branch Protection (Hard Rule)
 
-Apply equivalent settings in Gitea, GitHub, or GitLab:
+Apply equivalent settings in Gitea, GitHub, or GitLab, targeting the project's integration trunk
+(the branch its `.mosaic/repo.json` declares under `integration_trunk`; default `main` — see
+`CONSTITUTION.md` Hard Gates):
 
-1. Protect `main` from direct pushes.
-2. Require pull requests to merge into `main`.
+1. Protect the integration trunk from direct pushes.
+2. Require pull requests to merge into the integration trunk.
 3. Require required CI/status checks to pass before merge.
 4. Require code review approval before merge.
-5. Allow **squash merge only** for PRs into `main` (disable merge commits and rebase merges for `main`).
+5. Allow **squash merge only** for PRs into the integration trunk (disable merge commits and rebase merges for it).
 
 This enforces one merge strategy across human and agent workflows.
 
@@ -513,9 +515,9 @@ After bootstrapping, verify:
 - [ ] Git labels created (epic, feature, bug, task, etc.)
 - [ ] Initial pre-MVP milestone created (0.0.1)
 - [ ] MVP milestone reserved for release (0.1.0)
-- [ ] `main` is protected from direct pushes
-- [ ] PRs into `main` are required
-- [ ] Merge method for `main` is squash-only
+- [ ] The integration trunk is protected from direct pushes
+- [ ] PRs into the integration trunk are required
+- [ ] Merge method for the integration trunk is squash-only
 - [ ] Quality gates run successfully
 - [ ] `.env.example` exists (if project uses env vars)
 - [ ] CI/CD pipeline configured (if using Woodpecker/GitHub Actions)

packages/mosaic/framework/guides/CI-CD-PIPELINES.md

@@ -4,6 +4,11 @@
 
 ## Overview
 
+> **Integration trunk:** the YAML examples in this guide use the default integration trunk `main`
+> in branch conditions and version rules. A project that declares a different trunk in its
+> `.mosaic/repo.json` under `integration_trunk` (see `CONSTITUTION.md` Hard Gates) substitutes its
+> declared trunk wherever `main` appears as the trunk branch.
+
 This guide covers the canonical CI/CD pattern used across projects. The pipeline runs in Woodpecker CI and follows this flow:
 
 ```
@@ -865,7 +870,7 @@ steps:
    ```yaml
    image: git.example.com/org/service@${IMAGE_DIGEST}
    ```
-7. **Test on a short-lived non-main branch first** — open a PR and verify quality gates before merging to `main`
+7. **Test on a short-lived non-trunk branch first** — open a PR and verify quality gates before merging to the integration trunk
 8. **Verify images appear** in Gitea Packages tab after successful pipeline
 
 ## Terminal-Green Full-Step Contract
@@ -906,7 +911,7 @@ For source-code delivery, completion is not allowed at "PR opened" stage.
 
 Required sequence:
 
-1. Merge PR to `main` (squash) via Mosaic wrapper.
+1. Merge PR to the integration trunk (squash) via Mosaic wrapper.
 2. Monitor CI to terminal status:
    ```bash
    ~/.config/mosaic/tools/git/pr-ci-wait.sh -n <PR_NUMBER>
@@ -1112,5 +1117,5 @@ If a project currently uses Verdaccio (e.g., U-Connect at `npm.uscllc.net`), fol
 
 ### Pipeline runs Docker builds on pull requests
 
-- Verify `when` clause on Docker build steps restricts to `branch: [main]`
+- Verify `when` clause on Docker build steps restricts to the integration trunk (`branch: [main]` by default)
 - Pull requests should only run quality gates, not build/push images

packages/mosaic/framework/guides/CODE-REVIEW.md

@@ -10,13 +10,97 @@ If implementation diverges from `docs/PRD.md` or `docs/PRD.json` without PRD upd
 
 Merge strategy enforcement (HARD RULE):
 
-- PR target for delivery is `main`.
-- Direct pushes to `main` are prohibited.
-- Merge to `main` MUST be squash-only.
+- The integration trunk is the branch the project's `.mosaic/repo.json` declares under `integration_trunk` (default: `main`) — see `CONSTITUTION.md` Hard Gates.
+- PR target for delivery is the integration trunk.
+- Direct pushes to the integration trunk are prohibited.
+- Merge to the integration trunk MUST be squash-only.
 - Use `~/.config/mosaic/tools/git/pr-merge.sh -n {PR_NUMBER} -m squash --expect-head {approved_full_sha}` (or PowerShell equivalent).
 
+An estate MAY carry a documented exception for a repository whose gates are commit hooks rather
+than review. Such an exception belongs in that estate's own working copy of this guide, is
+scoped to the named repository, and is never precedent for a second one.
+
+**Do not use `pr-review.sh` or `issue-comment.sh` to post a verdict** (mosaicstack#1280). Post
+through a direct authenticated API call as your own seat, or hand the verdict to the requesting
+seat. Handing it over is a legitimate delivery path, not a fallback.
+
+## Evidence Discipline (applies to every finding)
+
+The checklist below says what to look at. This section says when you are allowed to believe what
+you saw. Every rule here was earned by a wrong conclusion that reached a report.
+
+1. **A finding is a claim about behavior.** State the failing input, the path taken, and the
+   wrong result. "This looks fragile" is not a finding.
+2. **A green check is not a result until you have shown it could go red.** Run the control. A
+   `0`, an empty result, or a column of identical values with no failing counterpart is a
+   non-result.
+3. **Measurement and explanation are separate sentences.** Report the command and its output,
+   then, as its own sentence, what you think it means.
+4. **Never widen the case you measured.** If you checked one path, the finding covers one path.
+5. **Reproduce a reported failure before recording it, and say which tree you measured.** Two
+   correct measurements of two different trees disagree without either being wrong.
+6. **Verify by content on the ref that ships**, never by ancestry of a local sha. A rebase mints
+   new shas; a commit being an ancestor of something local proves nothing about the remote.
+   Compare by digest against `origin/<branch>`.
+7. **Confidence is part of the finding.** "I could not reproduce this" is a usable review
+   comment. A confident guess is not.
+8. **Author is not reviewer** (Gate-16). Do not review your own work, or work you shaped closely
+   enough to be a co-author of. Say so and hand it back.
+
+### Measuring a shell suite
+
+Each of these produced a wrong conclusion before it was written down.
+
+9. **`cmd | tail; echo rc=$?` reports `tail`'s exit code, not `cmd`'s.** It reads as a pass when
+   the command failed. Redirect to a file and check `rc` directly, or use `${PIPESTATUS[0]}`.
+10. **Under `set -o pipefail`, a missed glob makes `ls` exit 2**, the pipeline inherits it, and
+    `set -e` kills the run. Iterate a glob with a `for` loop and an `-e` test instead of piping
+    `ls`.
+11. **A suite that exits nonzero with ZERO output is an environment question, not a defect in
+    the code under review.** The usual cause is a sourced dependency that is absent, so `set -e`
+    kills the first case before anything prints. Extract whole tool trees — `tools/git` alone is
+    missing `tools/_lib/credentials.sh`. Isolate the variable and prove it by adding only that
+    back.
+12. **`git -C <dir>` in a directory that is not itself a repo answers from the enclosing repo.**
+    A scratch tree under `~/.mosaic` reports `~/.mosaic`'s HEAD, not the PR's, and every
+    conclusion drawn from it describes the wrong tree. Confirm `git rev-parse --show-toplevel`
+    is the tree you think it is before trusting any git output.
+
+13. **Run the repository's PINNED tool version.** `npx <tool>` resolves a local `node_modules`
+    install when one is present and fetches the latest release when one is not, so the same
+    command answers differently depending on where it ran. A reviewer measuring in a fresh clone
+    or a detached worktree — which is exactly where reviewers measure — has no `node_modules` and
+    silently gets the latest release instead of the pinned one. Measured on mosaicstack#1313: the
+    lockfile pins prettier 3.8.1, under which three guides pass; a version-less `npx` in a
+    worktree resolved 3.9.6, under which the same three fail; and 3.0.0, the floor of the declared
+    `^3.0.0` range, fails a different one. Three versions, three verdicts, identical bytes. Use
+    `node_modules/.bin/<tool>`, or name the version the lockfile pins.
+14. **A formatter or linter declared as a range is a dated verdict, not a fact.** If a lockfile
+    pins it, the gate is reproducible today and will disagree with itself the day the pin moves.
+    Report a formatting failure with the version that produced it, always.
+
+### Feedback Categories
+
+- **Blocker**: must fix before merge (security, bugs, test failures)
+- **Should Fix**: important but not blocking (code quality, minor issues)
+- **Suggestion**: optional improvement (style preference, nice-to-have)
+- **Question**: seeking clarification
+
 ## Review Checklist
 
+Reviewer seats split this checklist by class rather than duplicating it. A seat reviews its own
+sections in full and may raise anything it notices outside them as a Suggestion, never as a
+Blocker on someone else's ground.
+
+| Reviewer class   | Owns                                                                                                    |
+| ---------------- | ------------------------------------------------------------------------------------------------------- |
+| `rev-code-*`     | 1 Correctness, 3 Testing, 4 Code Quality, 4a TypeScript, 5 Documentation, 6 Performance, 7 Dependencies |
+| `rev-security-*` | 2 Security, 2a OWASP                                                                                    |
+
+Where two seats of the same class review the same change, they review independently and compare
+after. A second seat that reads the first seat's findings before measuring is a proofreader, not
+a second opinion.
+
 ### 1. Correctness
 
 - [ ] Code does what the issue/PR description says
@@ -53,7 +137,7 @@ Merge strategy enforcement (HARD RULE):
 - [ ] Tests cover happy path AND error cases
 - [ ] Situational tests cover all impacted change surfaces (primary gate)
 - [ ] Tests validate required behavior/outcomes, not only internal implementation details
-- [ ] TDD was applied when required by `~/.config/mosaic/guides/QA-TESTING.md`
+- [ ] TDD was applied when required by `guides/QA-TESTING.md`
 - [ ] Coverage meets 85% minimum
 - [ ] Tests are readable and maintainable
 - [ ] No flaky tests introduced
@@ -82,7 +166,7 @@ Merge strategy enforcement (HARD RULE):
 ### 5. Documentation
 
 - [ ] Complex logic has explanatory comments
-- [ ] Required docs updated per `~/.config/mosaic/guides/DOCUMENTATION.md`
+- [ ] Required docs updated per `guides/DOCUMENTATION.md`
 - [ ] Public APIs are documented
 - [ ] Private/internal APIs are documented
 - [ ] API input/output schemas are documented
@@ -114,8 +198,8 @@ Use `~/.config/mosaic/templates/docs/DOCUMENTATION-CHECKLIST.md` whenever code/A
 # List the issue being addressed
 ~/.config/mosaic/tools/git/issue-list.sh -i {issue-number}
 
-# View the changes
-git diff main...HEAD
+# View the changes (diff against the integration trunk; default: main)
+git diff {integration_trunk}...HEAD
 ```
 
 ### Providing Feedback
@@ -126,13 +210,6 @@ git diff main...HEAD
 - Distinguish between blocking issues and suggestions
 - Be constructive, not critical of the person
 
-### Feedback Categories
-
-- **Blocker**: Must fix before merge (security, bugs, test failures)
-- **Should Fix**: Important but not blocking (code quality, minor issues)
-- **Suggestion**: Optional improvements (style preferences, nice-to-haves)
-- **Question**: Seeking clarification
-
 ### Review Comment Format
 
 ```
@@ -151,4 +228,4 @@ This pattern appears in 3 places. A shared helper would reduce duplication.
 2. If changes requested, assign back to author
 3. If approved, note approval in issue comments
 4. For merges, ensure CI passes first
-5. Merge PR to `main` with squash strategy only
+5. Merge PR to the integration trunk with squash strategy only

packages/mosaic/framework/guides/E2E-DELIVERY.md

@@ -78,7 +78,7 @@ For implementation work, you MUST run this cycle in order:
 7. `commit` - commit only when the logical unit passes tests and review.
 8. `pre-push queue guard` - before pushing, wait for running/queued project pipelines to clear: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push`.
 9. `push` - push immediately after queue guard passes.
-10. `PR integration` - if external git provider is available, create/update PR to `main` and merge with required strategy via Mosaic wrappers.
+10. `PR integration` - if external git provider is available, create/update PR to the integration trunk (the project's declared trunk, default `main`) and merge with required strategy via Mosaic wrappers.
 11. `pre-merge queue guard` - before merging PR, wait for running/queued project pipelines on the exact PR head to clear: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B <PR_HEAD_BRANCH> -R <PR_HEAD_OWNER/REPO> --sha <PR_HEAD_FULL_SHA>`.
 12. `CI/pipeline verification` - wait for terminal CI status and require green before completion (`~/.config/mosaic/tools/git/pr-ci-wait.sh` for PR-based workflow).
 13. `issue closure` - close linked external issue (or close internal `docs/TASKS.md` task ref when provider is unavailable).
@@ -199,7 +199,7 @@ Before running this checklist, pause and self-interrogate: did I fulfill the use
 10. No unresolved blocker hidden.
 11. If deployment is in scope, deployment target, release version, and post-deploy verification evidence are documented.
 12. `docs/TASKS.md` status and issue/internal references are updated to match delivered work.
-13. If source code changed and external provider is available: PR merged to `main` (squash), with merge evidence recorded.
+13. If source code changed and external provider is available: PR merged to the integration trunk (squash), with merge evidence recorded.
 14. CI/pipeline status is terminal green for the merged PR/head commit.
 15. Linked external issue is closed (or internal task ref is closed when no provider exists).
 16. If any of items 13-15 fail due access/tooling, report `blocked` with exact failed wrapper command and do not claim completion.

packages/mosaic/framework/guides/FLEET-COMMS.md

@@ -0,0 +1,86 @@
+# Fleet Comms Guide
+
+How one seat reaches another on a host. The mechanism is the framework's; the sessions and
+sockets are per-host, so measure yours rather than trusting an example.
+
+`mosaic <runtime>` would normally inject the addressing block from the roster. Where the composer
+is unavailable, or where the roster is stale, this guide is the substitute.
+
+## Measure the fleet; do not trust the roster
+
+`fleet/roster.yaml` is a declaration of intent, not an observation. It routinely names a socket
+that was never created, lists seats that are not running, and omits seats that are — this was
+all three have been observed true at once on a live host. Find out what is actually
+up before addressing anyone:
+
+```bash
+tmux list-sessions
+tmux list-panes -a -F '#{session_name}  #{pane_current_command}  #{pane_current_path}'
+```
+
+The pane command tells you the runtime. A pane showing `bash` is an idle shell with no agent
+attached — a send there lands in a shell prompt and is not read by anyone.
+
+Use the **default socket**. Do not pass `-L mosaic-fleet` on the strength of the roster.
+
+## Sending
+
+```bash
+~/.config/mosaic/tools/tmux/agent-send.sh -s <dst_session> -C <class> -m "<message>"
+```
+
+`-s` also accepts `session:window.pane`. `-f <file>` sends a file body; stdin works too.
+
+### Classes
+
+`-C` takes exactly one of these. Anything else exits 3.
+
+| Class          | Use for                                                  |
+| -------------- | -------------------------------------------------------- |
+| `terminal-log` | log only; never needs the agent's attention              |
+| `actionable`   | a decision, blocker, gate, or question needing an answer |
+| `human`        | relayed from a human operator                            |
+| `reaction`     | an ack or acknowledgement token                          |
+| `digest`       | machine wake, coalescible                                |
+
+An absent class is treated as `actionable` by consumers, which is the fail-safe direction. Prefer
+naming it anyway.
+
+### Addressing preamble
+
+The wire format is `[<src> -> <dst> class=<class>] <body>`. Flip it when you reply — the tool
+sends, it does not auto-reply.
+
+### Exit codes
+
+| rc  | Meaning                                        |
+| --- | ---------------------------------------------- |
+| 0   | delivered or queued                            |
+| 1   | target session not found                       |
+| 2   | text reached the pane but is **still a draft** |
+| 3   | usage error (bad class, missing `-s`)          |
+
+**Never retry on rc=2.** The message is in the target pane; retrying double-sends it. Confirm
+instead:
+
+```bash
+tmux capture-pane -p -t <session>:0.0 | tail -20
+```
+
+rc=2 is the normal result when the target is an idle pi seat.
+
+## Durable comms
+
+tmux delivery is host-local and does not survive a pane. Anything that must outlive the session
+goes through the estate's durable comms protocol — a committed `comms/` tree in an estate repo,
+with its own README. Use it for cross-host messages, verdicts, and anything a later session needs
+to find.
+
+## Handing work across seats
+
+1. **A verdict handed to the requesting seat is a legitimate delivery path**, and the required one
+   for anything `pr-review.sh` would otherwise post (see `guides/CODE-REVIEW.md`).
+2. **Address the seat, not the runtime.** A seat name is a session name; whether it runs claude,
+   pi or codex is not the sender's business.
+3. **Say what you measured, not just what you concluded** — the receiving seat cannot see your
+   terminal.

packages/mosaic/framework/guides/INFRASTRUCTURE.md

@@ -219,7 +219,7 @@ Use the Cloudflare tools for any DNS configuration: pointing domains at services
 
 # Update an existing record (get record ID from record-list first)
 ~/.config/mosaic/tools/cloudflare/record-update.sh \
-  -z example.com -r <record-id> -t A -n myapp -c 10.0.0.5 -p
+  -z example.com -r <record-id> -t A -n myapp -c 192.0.2.5 -p
 ```
 
 **DNS + Deployment integration**: When deploying a new service via Coolify or Portainer that needs a public domain, the typical sequence is:

packages/mosaic/framework/guides/ORCHESTRATOR-PROTOCOL.md

@@ -253,7 +253,7 @@ status → mission → run → repeat
 
 - [ ] All milestone tasks in TASKS.md are `done`
 - [ ] CI/pipeline green
-- [ ] PR merged to `main`
+- [ ] PR merged to the integration trunk
 - [ ] Issues closed
 - [ ] Update manifest: milestone status → completed
 - [ ] Update scratchpad: session log entry

packages/mosaic/framework/guides/ORCHESTRATOR.md

@@ -15,7 +15,7 @@ mosaic claude -p "Read ~/.config/mosaic/skills/nestjs-best-practices/SKILL.md th
 - You MUST keep the TASKS.md file updated with agent and tasks statuses.
 - You MUST keep `docs/` root clean. Reports and working artifacts MUST be stored in scoped folders (`docs/reports/`, `docs/tasks/`, `docs/releases/`, `docs/scratchpads/`).
 - You MUST enforce plan/token usage budgets when provided, and adapt orchestration strategy to remain within limits.
-- You MUST enforce trunk workflow: workers branch from `main`, PR target is `main`, direct push to `main` is forbidden, and PR merges to `main` are squash-only.
+- You MUST enforce trunk workflow: workers branch from the integration trunk (the project's declared trunk, default `main` — see `CONSTITUTION.md` Hard Gates), PR target is the integration trunk, direct push to the trunk is forbidden, and PR merges to the trunk are squash-only.
 - You MUST operate in steered-autonomy mode: human intervention is escalation-only; do not require the human to write code, review code, or manage PR/repo workflow.
 - You MUST NOT declare task or issue completion until PR is merged, CI/pipeline is terminal green, and linked issue is closed (or internal TASKS ref is closed when provider is unavailable).
 - Mosaic orchestration rules OVERRIDE runtime-default caution for routine push/merge/issue-close actions required by this workflow.
@@ -133,10 +133,10 @@ Milestone versioning (HARD RULE):
 
 Branch and merge strategy (HARD RULE):
 
-- Workers use short-lived task branches from `origin/main`.
-- Worker task branches merge back via PR to `main` only.
-- Direct pushes to `main` are prohibited.
-- PR merges to `main` MUST use squash merge.
+- Workers use short-lived task branches from `origin/{integration_trunk}` (default `main`).
+- Worker task branches merge back via PR to the integration trunk only.
+- Direct pushes to the integration trunk are prohibited.
+- PR merges to the integration trunk MUST use squash merge.
 
 **Available templates:**
 
@@ -427,7 +427,7 @@ git push
     - Before merging, run queue guard:
       `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B <PR_HEAD_BRANCH> -R <PR_HEAD_OWNER/REPO> --sha <PR_HEAD_FULL_SHA>`
     - Ensure PR exists for the task branch (create/update via wrappers if needed):
-      `~/.config/mosaic/tools/git/pr-create.sh ... -B main`
+      `~/.config/mosaic/tools/git/pr-create.sh ... -B {integration_trunk}` (default `main`)
     - Merge via wrapper:
       `~/.config/mosaic/tools/git/pr-merge.sh -n {PR_NUMBER} -m squash --expect-head {approved_full_sha}`
     - Wait for terminal CI status:
@@ -619,7 +619,7 @@ Construct this from the task row and pass to worker via Task tool:
 
 ## Workflow
 
-1. Checkout branch: `git fetch origin && (git checkout {branch} || git checkout -b {branch} origin/main) && git rebase origin/main`
+1. Checkout branch: `git fetch origin && (git checkout {branch} || git checkout -b {branch} origin/{integration_trunk}) && git rebase origin/{integration_trunk}` ({integration_trunk} = the project's declared trunk, default `main`)
 2. Read `docs/PRD.md` or `docs/PRD.json` and align implementation with PRD requirements
 3. Read the finding details from the report
 4. Implement the fix following existing code patterns
@@ -637,7 +637,7 @@ Do NOT leave lint warnings or errors for someone else to clean up. 6. Run REQUIR
 For issue/PR/milestone operations, use scripts (NOT raw tea/gh):
 
 - `~/.config/mosaic/tools/git/issue-view.sh -i {N}`
-- `~/.config/mosaic/tools/git/pr-create.sh -t "Title" -b "Desc" -B main`
+- `~/.config/mosaic/tools/git/pr-create.sh -t "Title" -b "Desc" -B {integration_trunk}`
 - Push: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B {task_branch}`
 - Merge: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B {pr_head_branch} -R {pr_head_owner/repo} --sha {pr_head_full_sha}`
 - `~/.config/mosaic/tools/git/pr-merge.sh -n {PR_NUMBER} -m squash --expect-head {approved_full_sha}`
@@ -994,13 +994,13 @@ mv docs/reports/qa-automation/pending/*failing-file* docs/reports/qa-automation/
 
 ---
 
-## Merge-to-Main Candidate Protocol (Container Deployments)
+## Merge-to-Trunk Candidate Protocol (Container Deployments)
 
-If deployment is in scope and container images are used, every merge to `main` MUST execute this protocol:
+If deployment is in scope and container images are used, every merge to the integration trunk MUST execute this protocol:
 
 1. Build and push immutable candidate image tags:
    - `sha-<shortsha>` (always)
-   - `v{base-version}-rc.{build}` (for `main` merges)
+   - `v{base-version}-rc.{build}` (for integration-trunk merges)
    - `testing` mutable pointer to the same digest
 2. Resolve and record the image digest for each service.
 3. Deploy by digest to testing environment (never deploy by mutable tag alone).

packages/mosaic/framework/guides/SEAT-IDENTITY.md

@@ -0,0 +1,133 @@
+# Seat Identity & Credentials Guide
+
+Every agent that touches a Mosaic-managed git host acts as a named seat with its own credential.
+This guide is how that works on a host, and what an agent must never do with it.
+
+The mechanism below is the framework's. The specific paths, seats and stores are per-host:
+measure yours before trusting any of them.
+
+## The rule
+
+**One seat, one identity, one token file.** A seat never borrows another seat's credential, never
+falls back to a shared owner account, and never carries a second copy of its own token. A second
+copy is drift, and drift surfaces as the stale copy returning 401 — which reads as a revoked
+token and sends whoever debugs it somewhere else entirely.
+
+A credential refusal is correct behavior, not a bug to route around. If git refuses with a
+fail-closed diagnostic, the fix is to provision or correct _your_ identity. Escalate; do not
+substitute.
+
+## How a credential is resolved
+
+Find the helper the way **git** does, not with `command -v`. Git runs whatever
+`credential.helper` names, and on a Mosaic host that is an absolute path — so a PATH lookup
+answers a different question and the two disagree the moment the PATH copy is removed. It was
+removed on hosts that have completed that migration.
+
+```bash
+git config --get-all credential.helper     # every helper, in the order git tries them
+```
+
+Git tries **each** configured helper in turn until one supplies a credential. A fail-closed
+helper supplies nothing, so a second helper configured behind it silently becomes the one that
+answers. When you care which binary serves a credential, read the whole list.
+Resolve all three forms git accepts — absolute path, `!command`, and a bare name looked up on
+PATH — not just the one your host happens to use.
+
+The helper resolves the identity in this order:
+
+1. `$MOSAIC_GIT_IDENTITY`
+2. `git config --get mosaic.gitIdentity`
+3. the username git supplied on stdin
+
+It maps the host to a store prefix — `git.mosaicstack.dev` to `gitea-mosaicstack`,
+`git.uscllc.com` to `gitea-usc`. Any other host is declined quietly with rc=0, which is not an
+error and raises no escalation.
+
+Then it chooses **one** of two stores, and reads exactly one file:
+
+```
+brain_home = ${MOSAIC_BRAIN_HOME:-$HOME/.mosaic}
+
+seat     — when $brain_home/fleet/agents/<identity>/ EXISTS
+           $brain_home/fleet/agents/<identity>/secrets/<prefix>-<identity>.token
+service  — otherwise
+           ~/.config/mosaic/secrets/gitea-tokens/<prefix>-<identity>.token
+```
+
+**There is no precedence between the two and no fallback from one to the other.** The existence
+of the seat directory decides it. A seat that has a directory and an empty slot fails closed; it
+does not reach the service store. That is the intended behavior — the alternative is an agent
+silently acting as somebody else.
+
+If the file is unreadable the helper **fails closed**: it refuses and writes a durable record to
+the escalation spool. It does not fall back to a shared account. The record is what exists — any
+alerting built on top of it is a separate, best-effort concern and is not performed by the helper,
+so do not wait for a notification that nothing sends. That fallback is what made
+`usc/uconnect#3084` unattributable, and it was removed deliberately.
+
+Verify the helper you actually have:
+
+```bash
+h=$(git config --get credential.helper)
+grep -c 'FAIL CLOSED'  "$h"    # expect >= 1
+grep -c 'fleet/agents' "$h"    # expect >= 1; 0 means it predates mosaicstack#1311
+```
+
+## Where a seat's token lives
+
+The seat slot is the **only** copy:
+
+```
+~/.mosaic/fleet/agents/<seat>/secrets/<prefix>-<seat>.token      real file, mode 600
+```
+
+The framework store at `~/.config/mosaic/secrets/gitea-tokens/` holds tokens for **service
+identities only** — identities with no seat directory. A seat's token does not belong there.
+
+Before mosaicstack#1311 the deployed helper knew only the service store, and seats were bridged
+with a symlink from the store into the slot. **Those bridges must be removed once a seat-aware helper is deployed, and must not be
+recreated.** Remove them only after the helper can reach the slot without them; the reverse order
+takes every seat offline. A symlink
+is not how a system finds a credential; the helper resolving the right store is.
+
+`.principal` and `.scopes` beside the token are grant records, not secrets. They are tracked. The
+`.token` never is.
+
+### Provisioning a new seat
+
+1. Create `~/.mosaic/fleet/agents/<seat>/secrets/` mode 700.
+2. Write `.principal` (the Gitea login) and `.scopes` (the granted scopes), mode 600.
+3. The estate operator mints the token into the seat slot, mode 600. Agents do not mint their
+   own, and do not ask another agent to mint one for them.
+4. Verify with an authenticated `GET /user` and confirm the returned login is the seat, **not the
+   minting account**. Record the date in `ENTITY.md`. Never record the value.
+
+There is no step that links the framework store to the slot. A seat-aware helper reads the slot
+directly; a store entry pointing at a slot is the bridge described in **Where a seat's token lives** above,
+and it is not part of provisioning.
+
+Until step 3, the seat is unminted and its git writes fail closed. That is the designed state and
+is safe to launch in — the seat is told at launch so it does not discover it mid-task.
+
+## Acting as yourself
+
+Name the identity on every invocation:
+
+```bash
+MOSAIC_GIT_IDENTITY=<seat> git push
+git -c user.name=<seat> -c user.email=<seat>@mosaicstack.dev commit -m "..."
+```
+
+**Never persist `git config mosaic.gitIdentity` inside a `~/src/stack` worktree.** Every worktree
+of that clone shares one `.git/config`, so a persisted identity there silently rewrites the
+identity of every other seat working in that clone. The per-invocation form has no exception.
+
+## Handling
+
+1. **Never print a token value.** Compare by SHA-256 digest, or write `<REDACTED>`.
+2. **Never stage a `.token`, `secrets.json`, or `ENTITY.md`.** Stage explicit paths and **never
+   `git add -A`** — `secrets/*.principal` and `secrets/*.scopes` are covered by no ignore rule.
+3. **Never place a token in an environment variable** in an interactive session. A `declare -x`
+   dump has leaked the whole environment to a terminal before.
+4. **No real credential or operator data on a sandbox VM, ever.**