feat(fleet): add local canary CLI

.gitignore

@@ -12,3 +12,6 @@ docs/reports/
 
 # Step-CA dev password — real file is gitignored; commit only the .example
 infra/step-ca/dev-password
+
+# Scratch dirs created by the framework git-wrapper shell test harnesses
+.mosaic-test-work/

docs/PRD.md

@@ -64,6 +64,7 @@ Jarvis (v0.2.0) is a self-hosted AI assistant with a Python FastAPI backend and
 21. `@mosaicstack/cli` — unified `mosaic` CLI
 22. Docker Compose deployment + bare-metal capability
 23. Agent log service — ingest, parse, tier, summarize agent interaction logs
+24. Local durable agent fleet canary — `mosaic fleet` / `mosaic agent` CLI for an isolated tmux-backed canary fleet using a named socket, with roster-driven local customization and rollback-safe verification
 
 ### Out of Scope (v0.1.0)
 

docs/guides/admin-guide.md

@@ -7,6 +7,7 @@
 3. [Provider Configuration](#provider-configuration)
 4. [MCP Server Configuration](#mcp-server-configuration)
 5. [Environment Variables Reference](#environment-variables-reference)
+6. [Local Fleet Canary](./fleet-local-canary.md)
 
 ---
 

docs/guides/dev-guide.md

@@ -9,6 +9,7 @@
 5. [Adding New MCP Tools](#adding-new-mcp-tools)
 6. [Database Schema and Migrations](#database-schema-and-migrations)
 7. [API Endpoint Reference](#api-endpoint-reference)
+8. [Local Fleet Canary](./fleet-local-canary.md)
 
 ---
 

docs/guides/fleet-local-canary.md

@@ -0,0 +1,112 @@
+# Local Fleet Canary
+
+The local fleet canary runs a small tmux-backed Mosaic agent fleet on an
+isolated tmux socket. The default socket is `mosaic-factory`; the commands do
+not use or stop the default tmux server.
+
+## Files
+
+Product-owned defaults:
+
+- `packages/mosaic/framework/fleet/roster.schema.json`
+- `packages/mosaic/framework/fleet/examples/minimal.yaml`
+- `packages/mosaic/framework/fleet/examples/local-canary.yaml`
+- `packages/mosaic/framework/systemd/user/mosaic-tmux-holder.service`
+- `packages/mosaic/framework/systemd/user/mosaic-agent@.service`
+- `packages/mosaic/framework/tools/fleet/start-agent-session.sh`
+- `packages/mosaic/framework/tools/tmux/agent-send.sh`
+- `packages/mosaic/framework/tools/tmux/send-message.sh`
+
+Site-owned local roster:
+
+```text
+~/.config/mosaic/fleet/roster.yaml
+```
+
+Do not put a host-specific full roster into product defaults. Start from an
+example and edit the local roster after `mosaic fleet init --write`.
+
+## Install
+
+Minimal canary:
+
+```bash
+mosaic fleet init --profile minimal --write
+# If a site-owned roster already exists, inspect it first; overwrite only explicitly:
+# mosaic fleet init --profile minimal --write --force
+mosaic fleet install-systemd
+systemctl --user daemon-reload
+mosaic fleet start
+mosaic fleet verify
+```
+
+Small dogfood roster:
+
+```bash
+mosaic fleet init --profile local-canary --write
+# Use --force only after preserving any site-owned roster changes.
+mosaic fleet install-systemd
+systemctl --user daemon-reload
+mosaic fleet start
+mosaic fleet status
+```
+
+## Agent Operations
+
+```bash
+mosaic agent roster
+mosaic agent status
+mosaic agent status canary-pi
+mosaic agent send canary-pi --message "status check"
+mosaic agent reset canary-pi --new
+mosaic agent tail canary-pi -n 80
+```
+
+These commands read the roster and target the configured tmux socket. The
+generated systemd agent services use `start-agent-session.sh`; message delivery
+uses the tmux send tools with `-L mosaic-factory`.
+
+## Verification
+
+Use these checks before expanding the roster:
+
+```bash
+tmux -L mosaic-factory ls
+tmux ls
+mosaic fleet verify
+systemctl --user status mosaic-tmux-holder.service
+```
+
+Expected results:
+
+- `tmux -L mosaic-factory ls` shows `_holder` and roster agent sessions.
+- `tmux ls` shows only the default tmux server sessions and is not changed by
+  fleet start/stop operations.
+- `mosaic fleet verify` checks exact session targets on the isolated socket.
+
+## Rollback
+
+Stop the local canary:
+
+```bash
+mosaic fleet stop
+systemctl --user disable mosaic-agent@canary-pi.service
+systemctl --user disable mosaic-tmux-holder.service
+systemctl --user daemon-reload
+```
+
+For a full local cleanup of generated canary files:
+
+```bash
+rm -f ~/.config/systemd/user/mosaic-agent@.service
+rm -f ~/.config/systemd/user/mosaic-tmux-holder.service
+rm -rf ~/.config/mosaic/fleet
+rm -rf ~/.config/mosaic/tools/fleet
+```
+
+This rollback leaves the default tmux server untouched. If a canary session is
+still present after service stop, remove only the isolated socket server:
+
+```bash
+tmux -L mosaic-factory kill-server
+```

docs/guides/user-guide.md

@@ -10,6 +10,7 @@
 6. [CLI Usage](#cli-usage)
 7. [Sub-package Commands](#sub-package-commands)
 8. [Telemetry](#telemetry)
+9. [Local Fleet Canary](./fleet-local-canary.md)
 
 ---
 

docs/scratchpads/2026-06-20-fleet-cli-local-canary.md

@@ -0,0 +1,52 @@
+# Fleet CLI Local Canary Dogfood — 2026-06-20
+
+## Objective
+
+Move the durable tmux fleet PoC into a functional local canary on this server. This is **not** production deployment. It is a canary/dogfood path for a small local agent fleet using an isolated tmux socket.
+
+## Issue
+
+- Gitea issue: #562 — `feat(fleet): local CLI canary dogfood`
+
+## Scope
+
+Implement enough product surface to use the fleet locally:
+
+- `mosaic fleet init/install/start/stop/restart/status/verify`
+- `mosaic agent roster/status/send/reset/tail`
+- roster schema and examples
+- local canary docs and rollback instructions
+- tests for CLI behavior where practical
+- canary verification on named tmux socket `mosaic-factory`
+
+## Non-goals
+
+- No production rollout.
+- No migration of existing default tmux sessions.
+- No image build/deploy work.
+- No hardcoded USC/local roster as product default.
+
+## Acceptance Criteria
+
+- CLI can initialize a minimal roster outside product defaults.
+- CLI can install user systemd units and fleet helper scripts to a configurable Mosaic home.
+- CLI can start/stop/status/verify a canary fleet using `mosaic-factory`.
+- `mosaic agent send` uses existing named-socket/exact-target tmux tooling.
+- `mosaic agent reset` targets only the named agent session on the named socket.
+- Verification proves default tmux sessions remain untouched.
+- Baseline repo gates pass.
+- PR CI is green before merge.
+- Local canary evidence is captured after merge/install.
+
+## Budget / Routing
+
+- Agent: codex preferred.
+- Estimate: 25K-40K tokens.
+- Worker owns implementation/tests/docs in branch `feat/fleet-cli-local-canary`.
+- Orchestrator owns `docs/TASKS.md`, issue/PR/merge, and local canary install verification.
+
+## Progress
+
+- 2026-06-20: #557 PoC primitives merged to `main` as `45e2c2a`.
+- 2026-06-20: issue #562 created for local CLI canary dogfood.
+- 2026-06-20: worktree created at `/home/jarvis/src/mosaicstack-stack-worktrees/fleet-cli-local-canary`.

docs/scratchpads/559-560-wrapper-eval-login-20260620.md

@@ -0,0 +1,87 @@
+# Wrapper hardening fold-in: #559 (eval removal) + #560 (host-derived login)
+
+**Branch:** `fix/wrapper-hardening-tls-credpath-cicwait` (PR #551)
+**Worker:** coderlite0 (Sonnet lane) · coordinated by mos-claude
+**Date:** 2026-06-20
+**Scope:** `packages/mosaic/framework/tools/git/*.sh` only
+
+## What the issues asked for vs. what was already landed
+
+Both issues were largely satisfied by prior merged work; this fold-in closes the
+remaining gaps (regression tests + a loud diagnostic + one residual word-split site)
+rather than re-implementing finished functionality.
+
+### #559 — remove `eval` from issue-create.sh (and siblings)
+
+- `eval`-based command construction was already removed across the wrapper surface
+  (landed in #549). A full scan of `tools/git/*.sh` finds **zero** `eval` usages.
+- `issue-create.sh`, `pr-create.sh`, `issue-edit.sh`, `issue-assign.sh` already build
+  their `tea`/`gh` invocations as argv arrays (`CMD=(...)`, `"${CMD[@]}"`), so Markdown
+  bodies pass through verbatim.
+- **Residual found & fixed:** `issue-comment.sh` still used unquoted
+  `$(get_gitea_repo_args)` word-splitting (the comment body itself was already safely
+  quoted, so no injection bug — but it was the inconsistent, fragile pattern #559 targets,
+  and it failed silently when no login resolved). Converted to an argv array with an
+  explicit, loud login-resolution error.
+- **Added regression test:** `test-issue-create-body-safety.sh` — feeds a hostile
+  Markdown body (`$(touch SENTINEL)`, backticks, single/double quotes, `$HOME`/`${PATH}`,
+  pipes/`&&`/`;`) through `issue-create.sh` and asserts (1) no command substitution
+  executes (sentinel file never created) and (2) the `--description` `tea` receives is
+  byte-for-byte the original body.
+
+### #560 — auto-detect Gitea `--login` from repo origin host
+
+- Centralized host→login resolution already exists in `detect-platform.sh`
+  (`get_gitea_login_for_host` → `find_tea_login_for_host`, matching `urlparse(url).hostname`).
+  Every wrapper routes through it (or `get_gitea_login` / `get_gitea_login_for_repo_override`);
+  **no wrapper hardcodes `${GITEA_LOGIN:-mosaicstack}`**. Explicit `GITEA_LOGIN` wins only
+  when it matches the host (`tea_login_matches_host`), so stale overrides are rejected.
+- **Gap fixed — silent failure → loud diagnostic:** the failure path of
+  `get_gitea_login_for_host` returned non-zero with no message. Added
+  `print_gitea_login_diagnostic`, emitted to **stderr** on resolution failure: names the
+  unresolved host, lists available tea logins (name + host), and gives the `GITEA_LOGIN`
+  override + `tea login add` fix. Stderr-only, so it never contaminates stdout (the
+  resolved login name) or the log-grep assertions in the existing harnesses. Callers with
+  an API fallback (pr-merge, issue-close, pr-create, issue-create) still follow with their
+  own "using API fallback" line, giving a clear "no login → fallback" trail.
+- **Extended test:** `test-gitea-login-resolution.sh` now also asserts (a) the loud
+  diagnostic fires and lists available logins for an unresolved host, (b) login is derived
+  from origin host for **both** instances (mosaicstack + usc) via a scoped second `tea`
+  mock, and (c) a valid `GITEA_LOGIN` override is honored. The scoped mock keeps the
+  existing API-fallback assertions (which require mosaicstack to have _no_ tea login) valid.
+
+## Files changed (wrapper surface only)
+
+- `detect-platform.sh` — add `print_gitea_login_diagnostic`; call it on the
+  `get_gitea_login_for_host` failure path.
+- `issue-comment.sh` — argv array + loud login-resolution error (was unquoted
+  `$(get_gitea_repo_args)`).
+- `test-issue-create-body-safety.sh` — **new** (#559 regression).
+- `test-gitea-login-resolution.sh` — extended (#560 diagnostic + both-host + override).
+
+## Verification
+
+All wrapper harnesses pass locally:
+
+- `test-issue-create-body-safety.sh` — PASS
+- `test-gitea-login-resolution.sh` — PASS
+- `test-pr-merge-gitea-empty-uid.sh` — PASS
+- `test-pr-metadata-gitea.sh` — PASS
+- `test-lane-brief-pr-linkage.sh` — PASS
+
+## Open items flagged to mos-claude (orchestrator decisions)
+
+1. **CHANGELOG absent.** The task said "update CHANGELOG (append-only), keep the existing
+   #550/#551 entry." No CHANGELOG file exists anywhere in the repo, and #550/#551 are not
+   recorded in one. **ASSUMPTION:** documenting #559/#560 in this scratchpad + the PR
+   description (`Closes #559 Closes #560`) follows the repo's actual convention
+   (`docs/scratchpads/`). Did not invent a new CHANGELOG structure.
+2. **`docs/TASKS.md` is orchestrator single-writer.** It carries a "Workers read but never
+   modify" banner. As a worker I did **not** edit it; task tracking is via the linked Gitea
+   issues #559/#560 + this scratchpad. Orchestrator may add a rollup row if desired.
+3. **Wrapper `test-*.sh` are not CI-wired.** `.woodpecker/ci.yml` runs `pnpm
+typecheck/lint/format:check/test` (`turbo run test`); the framework dir has no
+   `package.json`, so these shell harnesses run **locally/manually only** — they do not gate
+   the PR in Woodpecker. **ASSUMPTION:** out of scope to wire a shell-test step into CI in
+   this PR (would broaden the diff beyond the wrapper surface). Flagging for a follow-up if
+   the fleet wants these gated.

docs/scratchpads/fleet-cli-local-canary-review-fixes.md

@@ -0,0 +1,54 @@
+# Fleet CLI Local Canary Review Fixes
+
+## Objective
+
+Fix only the two should-fix code review findings:
+
+1. Ensure `@mosaicstack/mosaic` declares `yaml` and lockfile state is current.
+2. Validate `mosaic agent status [agent]` against the fleet roster before constructing/running the tmux target.
+
+## Constraints
+
+- Do not modify `docs/TASKS.md`.
+- Leave changes uncommitted.
+- Run requested formatting and quality gates.
+
+## Plan
+
+1. Inspect manifest/lockfile state for `yaml`.
+2. Add failing regression test for `mosaic agent status typo`.
+3. Patch `registerFleetAgentCommands` status validation.
+4. Format touched files.
+5. Run requested tests, typecheck, and lint.
+6. Review final diff.
+
+## Progress
+
+- Loaded required repo/global/runtime instructions.
+- Confirmed `packages/mosaic/package.json` already declares `yaml`.
+- Confirmed `pnpm-lock.yaml` already has `packages/mosaic` importer entry for `yaml`.
+- Found `registerFleetAgentCommands` status path does not validate agent before building tmux target.
+
+## Verification
+
+- TDD red check: `pnpm --filter @mosaicstack/mosaic test -- src/commands/fleet.spec.ts`
+  failed before the production fix because `mosaic agent status typo` resolved instead of
+  rejecting.
+- Focused green check: `pnpm --filter @mosaicstack/mosaic test -- src/commands/fleet.spec.ts`
+  passed after adding roster validation.
+- Formatting: `pnpm exec prettier --write packages/mosaic/src/commands/fleet.ts packages/mosaic/src/commands/fleet.spec.ts docs/scratchpads/fleet-cli-local-canary-review-fixes.md`
+  completed with all files unchanged.
+- Requested tests: `pnpm --filter @mosaicstack/mosaic test -- src/commands/fleet.spec.ts src/cli-smoke.spec.ts`
+  passed with 36 tests.
+- Baseline typecheck: `pnpm typecheck` passed.
+- Baseline lint: `pnpm lint` passed.
+- Independent review: `~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted`
+  returned approve with 0 findings. Note: reviewer reported broader context inspection was limited
+  by its read-only sandbox, so review was based on the supplied diff.
+- `docs/TASKS.md` has no diff.
+
+## Risks
+
+- `docs/TASKS.md` intentionally untouched per user instruction.
+- Review finding 1 required no file edit: `packages/mosaic/package.json` already declares
+  `yaml`, and the `packages/mosaic` importer in `pnpm-lock.yaml` already includes `yaml`.

docs/scratchpads/t-a292e96f-gitea-pr-metadata.md

@@ -51,3 +51,48 @@ This repository currently has no root `CHANGELOG.md`; the scratchpad and `docs/T
   - PR #1908: `Dry run: would merge PR #1908 on git.uscllc.com with authenticated Gitea API fallback (base=main, method=squash).`
 - PR: `https://git.mosaicstack.dev/mosaicstack/stack/pulls/518`, branch `fix/t-a292e96f-gitea-pr-metadata`.
 - CI: Recent PR/push pipelines failed before clone/test execution due Woodpecker/Kubernetes PVC API timeout: `dial tcp 10.43.0.1:443: i/o timeout`. No repository test step executed in CI; local targeted verification above remains clean.
+
+## 2026-06-18 — PR #549 functional blocker remediation
+
+### Assignment
+
+Coordinator `mos-claude` assigned remediation for PR #549: fix `packages/mosaic/framework/tools/git/pr-metadata.sh` tmpfile cleanup where an `EXIT` trap references function-local `body_file` after the function returns inside `RAW=$(...)`, producing `body_file: unbound variable` on the authenticated success path and failing to clean up safely on early `set -e` exits.
+
+### Plan
+
+1. Add a non-vacuous Gitea test that exercises `curl_gitea_pull` with stubbed `curl` and `GITEA_TOKEN` instead of `MOSAIC_GITEA_PR_METADATA_RAW_FILE`.
+2. Prove the new test is RED against the current PR head.
+3. Replace the function-local `EXIT` cleanup with robust function-scoped tmpfile cleanup.
+4. Re-run targeted tests, `bash -n`, and review gates; commit and push branch only. Do not merge.
+
+### Constraints / assumptions
+
+- Do not modify prior injection/JSON fixes in `issue-edit`, `issue-assign`, or `milestone-create`.
+- Worker role: do not modify `docs/TASKS.md`; orchestrator remains the single writer.
+- Budget: no explicit token cap provided; keep scope to shell wrapper + targeted regression harness.
+
+### Remediation results
+
+- Rebased `fix/tooling-eval-injection-jq-json` onto `origin/main`; branch was already current.
+- Added a curl-stub regression path that does not use `MOSAIC_GITEA_PR_METADATA_RAW_FILE`, so it exercises `curl_gitea_pull` and its temp body file.
+- RED evidence: copied the new harness next to the pre-fix `HEAD` version of `pr-metadata.sh`; `MOSAIC_TEST_WORK_DIR=$PWD/.mosaic-test-work/pr-metadata-red-work .../test-pr-metadata-gitea.sh` failed with `body_file: unbound variable` on the curl success path.
+- Fix: replaced `EXIT` temp-file cleanup with a `RETURN`-scoped cleanup function that removes the body file while the function-local variable is still in scope, preserves the original return status, and clears the `RETURN` trap.
+- GREEN evidence:
+  - `MOSAIC_TEST_WORK_DIR=$PWD/.mosaic-test-work/pr-metadata-gitea-current packages/mosaic/framework/tools/git/test-pr-metadata-gitea.sh` passed.
+  - `bash -n packages/mosaic/framework/tools/git/pr-metadata.sh packages/mosaic/framework/tools/git/test-pr-metadata-gitea.sh` passed.
+  - `shellcheck -x -P . -e SC1090 packages/mosaic/framework/tools/git/pr-metadata.sh packages/mosaic/framework/tools/git/test-pr-metadata-gitea.sh` passed.
+
+### Review remediation
+
+- Codex review returned one should-fix: the early-exit test used `chmod 000`, which is not root-safe in container CI.
+- Remediation: changed the stubbed 2xx/cat-failure mode to replace the curl output with a broken symlink, which fails deterministically even as root and still validates cleanup via `rm -f -- "$body_file"`.
+
+### Second review remediation
+
+- Codex review found the 2xx `cat "$body_file"` read could be masked under command substitution semantics because the branch returned 0 unconditionally.
+- Remediation: both authenticated 2xx branches now use `cat "$body_file" || return $?` before returning success.
+- Strengthened the broken-symlink test to require the body-read failure and reject the later `Gitea API returned non-JSON` parse-failure path, so the test verifies the helper-level failure propagation rather than eventual downstream failure.
+
+### Final review gate
+
+- Codex review after remediation: approved (`0 blockers, 0 should-fix, 0 suggestions`).

packages/mosaic/framework/defaults/TOOLS.md

@@ -5,10 +5,39 @@ Tool suites live at `~/.config/mosaic/tools/<suite>/`. This is the index only.
 read it (or the relevant service guide) when your task actually touches that service.
 Project-specific tooling belongs in the project's `AGENTS.md`, not here.
 
+## ⚡ Most-used fleet tools (reach for these FIRST — don't hand-roll)
+
+You are a Mosaic fleet agent. These cover the highest-frequency cross-agent and git-provider
+tasks — use them before improvising with raw `tmux send-keys`, raw `tea`/`gh`/`glab`, or `curl`.
+
+**1. Message another agent** → `tools/tmux/agent-send.sh` (NOT raw `tmux send-keys`):
+
+```bash
+tools/tmux/agent-send.sh -s <target-session> -m "message"   # or -f <file> to send a file's contents
+```
+
+The coordinator session is `mos-claude` — send status, findings, and questions there.
+
+**2. Issues / PRs / milestones** → `tools/git/*.sh` wrappers (before raw `tea`/`gh`/`glab`):
+
+```bash
+tools/git/pr-create.sh ...   tools/git/issue-create.sh ...   tools/git/pr-merge.sh ...
+tools/git/ci-queue-wait.sh --purpose push|merge   # REQUIRED before any push/merge
+```
+
+**GITEA_LOGIN gotcha** — the wrappers default to login `mosaicstack`; on a USC repo that fails with
+`gitea / Error: GetUserByName ... not found`. Pick the login from the repo's `origin` host first:
+
+| origin host           | login                                    |
+| --------------------- | ---------------------------------------- |
+| `git.uscllc.com`      | `export GITEA_LOGIN=usc`                 |
+| `git.mosaicstack.dev` | default `mosaicstack` (no export needed) |
+
 ## Suites (use wrappers first)
 
 | Suite      | Path                                             | Purpose                                                                  |
 | ---------- | ------------------------------------------------ | ------------------------------------------------------------------------ |
+| tmux       | `tools/tmux/agent-send.sh`                       | inter-agent messaging (see "Most-used" above)                            |
 | git        | `tools/git/*.sh`                                 | issues, PRs, milestones, CI queue guard (platform-auto-detected)         |
 | woodpecker | `tools/woodpecker/*.sh`                          | CI pipelines (`-a mosaic`\|`usc`; match git remote host)                 |
 | portainer  | `tools/portainer/*.sh`                           | Docker Swarm stacks (status/redeploy/list)                               |

packages/mosaic/framework/fleet/README.md

@@ -0,0 +1,26 @@
+# Mosaic Fleet Rosters
+
+The local fleet canary uses a product-owned roster schema with site-owned roster
+files. Product examples live here; active local rosters should live outside the
+package, normally at:
+
+```text
+~/.config/mosaic/fleet/roster.yaml
+```
+
+The default tmux socket is `mosaic-factory` so fleet commands do not touch the
+default tmux server.
+
+## Examples
+
+- `examples/minimal.yaml` starts one local canary slot.
+- `examples/local-canary.yaml` starts a small generic dogfood fleet.
+
+Initialize a roster:
+
+```bash
+mosaic fleet init --profile minimal --write
+mosaic fleet install-systemd
+mosaic fleet start
+mosaic fleet verify
+```

packages/mosaic/framework/fleet/examples/local-canary.yaml

@@ -0,0 +1,27 @@
+version: 1
+transport: tmux
+tmux:
+  socket_name: mosaic-factory
+  holder_session: _holder
+defaults:
+  working_directory: ~/src
+runtimes:
+  claude:
+    reset_command: /clear
+  codex:
+    reset_command: /clear
+  pi:
+    reset_command: /new
+agents:
+  - name: lead
+    runtime: claude
+    class: orchestrator
+    persistent_persona: true
+  - name: coder0
+    runtime: codex
+    class: implementer
+    reset_between_tasks: true
+  - name: reviewer0
+    runtime: pi
+    class: reviewer
+    reset_between_tasks: true

packages/mosaic/framework/fleet/examples/minimal.yaml

@@ -0,0 +1,15 @@
+version: 1
+transport: tmux
+tmux:
+  socket_name: mosaic-factory
+  holder_session: _holder
+defaults:
+  working_directory: ~/src
+runtimes:
+  pi:
+    reset_command: /new
+agents:
+  - name: canary-pi
+    runtime: pi
+    class: canary
+    reset_between_tasks: true

packages/mosaic/framework/fleet/roster.schema.json

@@ -0,0 +1,118 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://mosaicstack.dev/schemas/fleet-roster.schema.json",
+  "title": "Mosaic Fleet Roster",
+  "type": "object",
+  "required": ["version", "transport", "agents"],
+  "additionalProperties": false,
+  "properties": {
+    "version": {
+      "const": 1
+    },
+    "transport": {
+      "const": "tmux"
+    },
+    "tmux": {
+      "type": "object",
+      "additionalProperties": false,
+      "properties": {
+        "socket_name": {
+          "type": "string",
+          "default": "mosaic-factory"
+        },
+        "socketName": {
+          "type": "string",
+          "default": "mosaic-factory"
+        },
+        "holder_session": {
+          "type": "string",
+          "default": "_holder"
+        },
+        "holderSession": {
+          "type": "string",
+          "default": "_holder"
+        }
+      }
+    },
+    "defaults": {
+      "type": "object",
+      "additionalProperties": false,
+      "properties": {
+        "working_directory": {
+          "type": "string",
+          "default": "~/src"
+        },
+        "workingDirectory": {
+          "type": "string",
+          "default": "~/src"
+        }
+      }
+    },
+    "runtimes": {
+      "type": "object",
+      "additionalProperties": {
+        "type": "object",
+        "additionalProperties": false,
+        "properties": {
+          "reset_command": {
+            "type": "string"
+          },
+          "resetCommand": {
+            "type": "string"
+          }
+        }
+      }
+    },
+    "agents": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": ["name", "runtime"],
+        "additionalProperties": false,
+        "properties": {
+          "name": {
+            "type": "string",
+            "pattern": "^[A-Za-z0-9_.-]+$"
+          },
+          "runtime": {
+            "type": "string"
+          },
+          "class": {
+            "type": "string"
+          },
+          "working_directory": {
+            "type": "string"
+          },
+          "workingDirectory": {
+            "type": "string"
+          },
+          "model_hint": {
+            "type": "string"
+          },
+          "modelHint": {
+            "type": "string"
+          },
+          "persistent_persona": {
+            "oneOf": [{ "type": "boolean" }, { "type": "string" }]
+          },
+          "persistentPersona": {
+            "oneOf": [{ "type": "boolean" }, { "type": "string" }]
+          },
+          "reset_between_tasks": {
+            "type": "boolean"
+          },
+          "resetBetweenTasks": {
+            "type": "boolean"
+          },
+          "kickstart_template": {
+            "type": "string"
+          },
+          "kickstartTemplate": {
+            "type": "string"
+          }
+        }
+      }
+    }
+  }
+}

packages/mosaic/framework/runtime/pi/RUNTIME.md

@@ -29,7 +29,21 @@ Pi supports `--models` for Ctrl+P model cycling during a session. Use cheaper mo
 
 ### Skills
 
-Mosaic skills are loaded natively via Pi's `--skill` flag. Skills are discovered from:
+By default the launcher starts Pi with `--no-skills` to keep startup context small, then
+force-loads a small set of fleet-critical skills via explicit `--skill` flags (an explicit
+`--skill` overrides `--no-skills` for that path). The default forced set is `mosaic-tools`
+(the must-use `~/.config/mosaic/tools/` cheatsheet: inter-agent messaging + git wrappers).
+
+Tune skill loading with environment variables:
+
+- `MOSAIC_PI_FORCE_SKILLS` — colon-separated skill dir names to force-load (default: `mosaic-tools`;
+  set to an empty string to disable force-loading). Missing skills are skipped silently.
+- `MOSAIC_PI_SKILL_MODE=all` — link every skill found in `~/.config/mosaic/{skills,skills-local}/`
+  (full catalog; larger context).
+- `MOSAIC_PI_SKILL_MODE=discover` — let Pi discover skills natively (no `--no-skills`), still
+  force-loading the fleet set on top.
+
+Skills are discovered from:
 
 - `~/.config/mosaic/skills/` (Mosaic global skills)
 - `~/.pi/agent/skills/` (Pi global skills)

packages/mosaic/framework/systemd/user/README.md

@@ -0,0 +1,57 @@
+# Mosaic tmux Fleet PoC
+
+This directory contains the first durable tmux-backed fleet primitives for the
+Mosaic software-factory model.
+
+The lifecycle model follows the organization-neutral AI Guide playbook
+`mosaicstack/aiguide:playbooks/tmux-fleet.md` (commit `2a0b0b5`): a dedicated
+holder owns the tmux server/socket; agent units join it and stop only their own
+exact-match session.
+
+## Layout
+
+- `mosaic-tmux-holder.service` — user-mode holder that owns the named tmux server.
+- `mosaic-agent@.service` — user-mode template for one reusable agent session.
+- `test-fleet-units.sh` — validates unit syntax and required relationships.
+
+The agent template calls:
+
+```text
+~/.config/mosaic/tools/fleet/start-agent-session.sh <agent-name>
+```
+
+which starts or reuses a tmux session on `MOSAIC_TMUX_SOCKET`.
+
+## Local customization
+
+Per-agent overrides live outside the package in:
+
+```text
+~/.config/mosaic/fleet/agents/<agent>.env
+```
+
+Example:
+
+```dotenv
+MOSAIC_TMUX_SOCKET=mosaic-factory
+MOSAIC_AGENT_RUNTIME=claude
+MOSAIC_AGENT_WORKDIR=/home/jarvis/src/mosaic-stack
+# Optional escape hatch for PoC/canary agents:
+# MOSAIC_AGENT_COMMAND=mosaic yolo claude
+```
+
+## Manual canary sequence
+
+```bash
+mkdir -p ~/.config/systemd/user ~/.config/mosaic/tools/fleet ~/.config/mosaic/fleet/agents
+cp packages/mosaic/framework/systemd/user/mosaic-*.service ~/.config/systemd/user/
+cp packages/mosaic/framework/tools/fleet/start-agent-session.sh ~/.config/mosaic/tools/fleet/
+chmod +x ~/.config/mosaic/tools/fleet/start-agent-session.sh
+systemctl --user daemon-reload
+systemctl --user start mosaic-tmux-holder.service
+systemctl --user start mosaic-agent@canary.service
+tmux -L mosaic-factory ls
+```
+
+Do not use `tmux kill-server` without `-L mosaic-factory`; this pattern is meant
+to avoid disturbing the user's default tmux server.

packages/mosaic/framework/systemd/user/mosaic-agent@.service

@@ -0,0 +1,20 @@
+[Unit]
+Description=Mosaic tmux fleet agent %i
+Documentation=https://git.mosaicstack.dev/mosaicstack/stack
+Requires=mosaic-tmux-holder.service
+After=mosaic-tmux-holder.service
+PartOf=mosaic-tmux-holder.service
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+Environment=MOSAIC_TMUX_SOCKET=mosaic-factory
+Environment=MOSAIC_AGENT_NAME=%i
+Environment=MOSAIC_AGENT_RUNTIME=pi
+Environment=MOSAIC_AGENT_WORKDIR=%h
+EnvironmentFile=-%h/.config/mosaic/fleet/agents/%i.env
+ExecStart=/bin/bash %h/.config/mosaic/tools/fleet/start-agent-session.sh %i
+ExecStop=-/bin/bash -lc 'tmux -L "${MOSAIC_TMUX_SOCKET:-mosaic-factory}" kill-session -t "=%i"'
+
+[Install]
+WantedBy=default.target

packages/mosaic/framework/systemd/user/mosaic-tmux-holder.service

@@ -0,0 +1,15 @@
+[Unit]
+Description=Mosaic tmux fleet holder
+Documentation=https://git.mosaicstack.dev/mosaicstack/stack
+After=default.target
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+Environment=MOSAIC_TMUX_SOCKET=mosaic-factory
+Environment=MOSAIC_TMUX_HOLDER=_holder
+ExecStart=/bin/bash -lc 'tmux -L "$MOSAIC_TMUX_SOCKET" has-session -t "=${MOSAIC_TMUX_HOLDER}:0.0" 2>/dev/null || tmux -L "$MOSAIC_TMUX_SOCKET" new-session -d -s "$MOSAIC_TMUX_HOLDER" "while true; do sleep 3600; done"'
+ExecStop=-/bin/bash -lc 'tmux -L "$MOSAIC_TMUX_SOCKET" kill-server'
+
+[Install]
+WantedBy=default.target

packages/mosaic/framework/systemd/user/test-fleet-units.sh

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR=$(cd -- "$(dirname -- "$0")" && pwd)
+HOLDER="$SCRIPT_DIR/mosaic-tmux-holder.service"
+AGENT="$SCRIPT_DIR/mosaic-agent@.service"
+
+fail() {
+  echo "FAIL: $*" >&2
+  exit 1
+}
+
+[ -f "$HOLDER" ] || fail "missing mosaic-tmux-holder.service"
+[ -f "$AGENT" ] || fail "missing mosaic-agent@.service"
+
+grep -qF 'ExecStart=' "$HOLDER" || fail "holder has no ExecStart"
+grep -qF 'tmux -L' "$HOLDER" || fail "holder does not use named tmux socket"
+grep -qF '_holder' "$HOLDER" || fail "holder session is not explicit"
+grep -qF 'Requires=mosaic-tmux-holder.service' "$AGENT" || fail "agent does not require holder"
+grep -qF 'start-agent-session.sh' "$AGENT" || fail "agent unit does not call start-agent-session.sh"
+grep -qF 'kill-session -t "=%i"' "$AGENT" || fail "agent stop does not exact-match its session"
+
+if command -v systemd-analyze >/dev/null 2>&1; then
+  systemd-analyze verify --user "$HOLDER" "$AGENT" >/tmp/mosaic-fleet-systemd-verify.log 2>&1 || {
+    cat /tmp/mosaic-fleet-systemd-verify.log >&2
+    fail "systemd-analyze verify failed"
+  }
+fi
+
+echo "ok - fleet systemd unit templates"

packages/mosaic/framework/templates/agent/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -58,7 +58,7 @@ ${QUALITY_GATES}
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -88,7 +88,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 ## Container Release Strategy (When Applicable)
 
@@ -138,8 +138,8 @@ When completing an orchestrated task:
 ### Post-Coding Review
 After implementing changes, code review is REQUIRED for any source-code modification.
 For orchestrated tasks, the orchestrator will run:
-1. **Codex code review** — `~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted`
-2. **Codex security review** — `~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted`
+1. **Codex code review** — `~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted`
+2. **Codex security review** — `~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted`
 3. If blockers/critical findings: remediation task created
 4. If clean: task marked done
 

packages/mosaic/framework/templates/agent/CLAUDE.md.template

@@ -135,7 +135,7 @@ ${QUALITY_GATES}
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -147,9 +147,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.
@@ -176,10 +176,10 @@ Run independent reviews:
 
 ```bash
 # Code quality review (Codex)
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
 
 # Security review (Codex)
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 **Fallback:** If Codex is unavailable, use Claude's built-in review skills.

packages/mosaic/framework/templates/agent/projects/django/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -68,7 +68,7 @@ ruff check . && mypy . && pytest tests/
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -97,7 +97,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 
 ## Container Release Strategy (When Applicable)
@@ -139,8 +139,8 @@ Use `${TASK_PREFIX}` for orchestrated tasks (e.g., `${TASK_PREFIX}-SEC-001`).
 ### Post-Coding Review
 After implementing changes, code review is REQUIRED for any source-code modification.
 For orchestrated tasks, the orchestrator will run:
-1. **Codex code review** — `~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted`
-2. **Codex security review** — `~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted`
+1. **Codex code review** — `~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted`
+2. **Codex security review** — `~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted`
 3. If blockers/critical findings: remediation task created
 4. If clean: task marked done
 

packages/mosaic/framework/templates/agent/projects/django/CLAUDE.md.template

@@ -159,10 +159,10 @@ Run independent reviews:
 
 ```bash
 # Code quality review (Codex)
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
 
 # Security review (Codex)
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 See `~/.config/mosaic/guides/CODE-REVIEW.md` for the full review checklist.
@@ -186,7 +186,7 @@ See `~/.config/mosaic/guides/DOCUMENTATION.md` for required documentation delive
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -198,9 +198,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.

packages/mosaic/framework/templates/agent/projects/nestjs-nextjs/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -72,7 +72,7 @@ pnpm typecheck && pnpm lint && pnpm test
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -101,7 +101,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 
 ## Container Release Strategy (When Applicable)
@@ -143,8 +143,8 @@ Use `${TASK_PREFIX}` for orchestrated tasks (e.g., `${TASK_PREFIX}-SEC-001`).
 ### Post-Coding Review
 After implementing changes, code review is REQUIRED for any source-code modification.
 For orchestrated tasks, the orchestrator will run:
-1. **Codex code review** — `~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted`
-2. **Codex security review** — `~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted`
+1. **Codex code review** — `~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted`
+2. **Codex security review** — `~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted`
 3. If blockers/critical findings: remediation task created
 4. If clean: task marked done
 

packages/mosaic/framework/templates/agent/projects/nestjs-nextjs/CLAUDE.md.template

@@ -191,10 +191,10 @@ Run independent reviews:
 
 ```bash
 # Code quality review (Codex)
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
 
 # Security review (Codex)
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 See `~/.config/mosaic/guides/CODE-REVIEW.md` for the full review checklist.
@@ -218,7 +218,7 @@ See `~/.config/mosaic/guides/DOCUMENTATION.md` for required documentation delive
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -230,9 +230,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.

packages/mosaic/framework/templates/agent/projects/python-fastapi/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -58,7 +58,7 @@ uv run ruff check src/ tests/ && uv run ruff format --check src/ && uv run mypy
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -87,7 +87,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 ## Container Release Strategy (When Applicable)
 

packages/mosaic/framework/templates/agent/projects/python-fastapi/CLAUDE.md.template

@@ -135,7 +135,7 @@ uv run ruff check src/ tests/ && uv run ruff format --check src/ && uv run mypy
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -146,9 +146,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.
@@ -171,8 +171,8 @@ If you modify source code, independent code review is REQUIRED before completion
 Run independent reviews:
 
 ```bash
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 See `~/.config/mosaic/guides/CODE-REVIEW.md` for the full review checklist.

packages/mosaic/framework/templates/agent/projects/python-library/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -55,7 +55,7 @@ uv run ruff check src/ tests/ && uv run ruff format --check src/ && uv run mypy
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -84,7 +84,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 ## Container Release Strategy (When Applicable)
 

packages/mosaic/framework/templates/agent/projects/python-library/CLAUDE.md.template

@@ -125,7 +125,7 @@ uv run ruff check src/ tests/ && uv run ruff format --check src/ && uv run mypy
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -136,9 +136,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.
@@ -161,8 +161,8 @@ If you modify source code, independent code review is REQUIRED before completion
 Run independent reviews:
 
 ```bash
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 See `~/.config/mosaic/guides/CODE-REVIEW.md` for the full review checklist.

packages/mosaic/framework/templates/agent/projects/typescript/AGENTS.md.template

@@ -9,8 +9,8 @@
 2. Do NOT ask for routine confirmation before required push/merge/issue-close/release/tag actions.
 3. Completion is forbidden at PR-open stage.
 4. Completion requires merged PR to `main` + terminal green CI + linked issue/internal task closed.
-5. Before push or merge, run queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
-6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/rails/git/*.sh`).
+5. Before push or merge, run queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
+6. For issue/PR/milestone operations, use Mosaic wrappers first (`~/.config/mosaic/tools/git/*.sh`).
 7. If any required wrapper command fails: report `blocked` with the exact failed wrapper command and stop.
 8. Do NOT stop at "PR created" and do NOT ask "should I merge?" for routine flow.
 
@@ -56,7 +56,7 @@ ${QUALITY_GATES}
 2. If external git provider is available (Gitea/GitHub/GitLab), create/update issue(s) before coding and map them in `docs/TASKS.md`.
 3. If no external provider is available, use internal refs in `docs/TASKS.md` (example: `TASKS:T1`).
 4. Keep `docs/TASKS.md` status in sync with actual progress until completion.
-5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
+5. For issue/PR/milestone actions, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first (no raw `gh`/`tea`/`glab` as first choice).
 6. If wrapper-driven merge/CI/issue-closure fails, report blocker with the exact failed wrapper command and stop (do not claim completion).
 
 ## Documentation Contract
@@ -85,7 +85,7 @@ Reference:
 5. Do not mark implementation complete until PR is merged.
 6. Do not mark implementation complete until CI/pipeline status is terminal green.
 7. Close linked issues/tasks only after merge + green CI.
-8. Before push or merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push|merge -B main`.
+8. Before push or merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push|merge -B main`.
 
 ## Container Release Strategy (When Applicable)
 

packages/mosaic/framework/templates/agent/projects/typescript/CLAUDE.md.template

@@ -122,7 +122,7 @@ ${QUALITY_GATES}
 ## Issue Tracking
 
 Use external git provider issues when available. If no external provider exists, `docs/TASKS.md` is the canonical tracker for tasks, milestones, and issue-equivalent work.
-For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/rails/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
+For issue/PR/milestone operations, detect platform and use `~/.config/mosaic/tools/git/*.sh` wrappers first; do not use raw `gh`/`tea`/`glab` as first choice.
 If wrapper-driven merge/CI/issue-closure fails, report blocker with exact failed wrapper command and stop.
 Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close the issue?" for routine delivery flow.
 
@@ -133,9 +133,9 @@ Do NOT stop at "PR created" and do NOT ask "should I merge?" or "should I close
 5. Ensure `docs/PRD.md` or `docs/PRD.json` exists and is current before coding.
 6. Create scratchpad: `docs/scratchpads/{task-id}-{short-name}.md` and include issue/internal ref.
 7. Update `docs/TASKS.md` status + issue/internal ref before coding.
-8. Before push, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose push -B main`.
+8. Before push, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose push -B main`.
 9. Open PR to `main` for delivery changes (no direct push to `main`).
-10. Before merge, run CI queue guard: `~/.config/mosaic/rails/git/ci-queue-wait.sh --purpose merge -B main`.
+10. Before merge, run CI queue guard: `~/.config/mosaic/tools/git/ci-queue-wait.sh --purpose merge -B main`.
 11. Merge PRs that pass required checks and review gates with squash strategy only.
 12. Reference issues/internal refs in commits (`Fixes #123`, `Refs #123`, or `Refs TASKS:T1`).
 13. Close issue/internal task only after testing and documentation gates pass, PR merge is complete, and CI/pipeline status is terminal green.
@@ -159,10 +159,10 @@ Run independent reviews:
 
 ```bash
 # Code quality review (Codex)
-~/.config/mosaic/rails/codex/codex-code-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-code-review.sh --uncommitted
 
 # Security review (Codex)
-~/.config/mosaic/rails/codex/codex-security-review.sh --uncommitted
+~/.config/mosaic/tools/codex/codex-security-review.sh --uncommitted
 ```
 
 **Fallback:** If Codex is unavailable, use Claude's built-in review skills.

packages/mosaic/framework/tools/_lib/credentials.sh

@@ -16,7 +16,12 @@
 # After loading, service-specific env vars are exported.
 # Run `load_credentials --help` for details.
 
-MOSAIC_CREDENTIALS_FILE="${MOSAIC_CREDENTIALS_FILE:-$HOME/src/jarvis-brain/credentials.json}"
+if [[ -z "${MOSAIC_CREDENTIALS_FILE:-}" ]]; then
+  for _cand in "$HOME/.config/mosaic/credentials.json" "$HOME/src/jarvis-brain/credentials.json"; do
+    if [[ -f "$_cand" ]]; then MOSAIC_CREDENTIALS_FILE="$_cand"; break; fi
+  done
+  : "${MOSAIC_CREDENTIALS_FILE:=$HOME/src/jarvis-brain/credentials.json}"
+fi
 
 _mosaic_require_jq() {
   if ! command -v jq &>/dev/null; then
@@ -34,6 +39,19 @@ _mosaic_read_cred() {
   jq -r "$jq_path // empty" "$MOSAIC_CREDENTIALS_FILE"
 }
 
+# Decide curl TLS flag for a target URL: validate public hosts (MITM matters on
+# WAN); allow self-signed only for private-network IP literals (trusted LAN) or an
+# explicit $MOSAIC_INSECURE_TLS opt-in. Echoes "-k" or "" (empty).
+_mosaic_tls_opt() {
+  local url="$1" host
+  [[ -n "${MOSAIC_INSECURE_TLS:-}" ]] && { echo "-k"; return; }
+  host=$(printf '%s' "$url" | sed -E 's#^[a-zA-Z]+://([^/:]+).*#\1#')
+  if [[ "$host" =~ ^(10\.|127\.|192\.168\.|172\.(1[6-9]|2[0-9]|3[01])\.) ]]; then
+    echo "-k"; return
+  fi
+  echo ""
+}
+
 # Sync Woodpecker credentials to ~/.woodpecker/<instance>.env
 # Only writes when values differ to avoid unnecessary disk writes.
 _mosaic_sync_woodpecker_env() {
@@ -261,7 +279,8 @@ mosaic_http() {
   local base_url="${4:-}"
 
   local response
-  response=$(curl -sk -w "\n%{http_code}" -X "$method" \
+  local _tls; _tls=$(_mosaic_tls_opt "${base_url}${endpoint}")
+  response=$(curl -sS $_tls -w "\n%{http_code}" -X "$method" \
     -H "$auth_header" \
     -H "Content-Type: application/json" \
     "${base_url}${endpoint}")
@@ -279,7 +298,8 @@ mosaic_http_post() {
   local base_url="${4:-}"
 
   local response
-  response=$(curl -sk -w "\n%{http_code}" -X POST \
+  local _tls; _tls=$(_mosaic_tls_opt "${base_url}${endpoint}")
+  response=$(curl -sS $_tls -w "\n%{http_code}" -X POST \
     -H "$auth_header" \
     -H "Content-Type: application/json" \
     -d "$data" \
@@ -297,7 +317,8 @@ mosaic_http_patch() {
   local base_url="${4:-}"
 
   local response
-  response=$(curl -sk -w "\n%{http_code}" -X PATCH \
+  local _tls; _tls=$(_mosaic_tls_opt "${base_url}${endpoint}")
+  response=$(curl -sS $_tls -w "\n%{http_code}" -X PATCH \
     -H "$auth_header" \
     -H "Content-Type: application/json" \
     -d "$data" \

packages/mosaic/framework/tools/fleet/start-agent-session.sh

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+AGENT_NAME=${1:-${MOSAIC_AGENT_NAME:-}}
+MOSAIC_TMUX_SOCKET=${MOSAIC_TMUX_SOCKET:-mosaic-factory}
+MOSAIC_AGENT_RUNTIME=${MOSAIC_AGENT_RUNTIME:-pi}
+MOSAIC_AGENT_WORKDIR=${MOSAIC_AGENT_WORKDIR:-$HOME}
+MOSAIC_AGENT_COMMAND=${MOSAIC_AGENT_COMMAND:-}
+
+if [ -z "$AGENT_NAME" ]; then
+  echo "ERROR: agent name argument or MOSAIC_AGENT_NAME is required" >&2
+  exit 64
+fi
+
+if ! command -v tmux >/dev/null 2>&1; then
+  echo "ERROR: tmux is required" >&2
+  exit 69
+fi
+
+if tmux -L "$MOSAIC_TMUX_SOCKET" has-session -t "=${AGENT_NAME}:0.0" 2>/dev/null; then
+  echo "Mosaic agent session already running: $AGENT_NAME on socket $MOSAIC_TMUX_SOCKET"
+  exit 0
+fi
+
+if [ -z "$MOSAIC_AGENT_COMMAND" ]; then
+  MOSAIC_AGENT_COMMAND="mosaic yolo $MOSAIC_AGENT_RUNTIME"
+fi
+
+mkdir -p "$MOSAIC_AGENT_WORKDIR"
+exec tmux -L "$MOSAIC_TMUX_SOCKET" new-session -d -s "$AGENT_NAME" -c "$MOSAIC_AGENT_WORKDIR" "$MOSAIC_AGENT_COMMAND"

packages/mosaic/framework/tools/fleet/test-start-agent-session.sh

@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR=$(cd -- "$(dirname -- "$0")" && pwd)
+START="$SCRIPT_DIR/start-agent-session.sh"
+SOCKET="mosaic-agent-test-$RANDOM-$$"
+AGENT="agent-$RANDOM"
+WORKDIR=$(mktemp -d)
+trap 'tmux -L "$SOCKET" kill-server >/dev/null 2>&1 || true; rm -rf "$WORKDIR"' EXIT
+
+fail() {
+  echo "FAIL: $*" >&2
+  exit 1
+}
+
+MOSAIC_TMUX_SOCKET="$SOCKET" \
+MOSAIC_AGENT_WORKDIR="$WORKDIR" \
+MOSAIC_AGENT_COMMAND='bash --noprofile --norc -i' \
+  "$START" "$AGENT"
+
+tmux -L "$SOCKET" has-session -t "=$AGENT:0.0" || fail "agent session was not created"
+actual_dir=$(tmux -L "$SOCKET" display-message -p -t "=$AGENT:0.0" '#{pane_current_path}')
+[ "$actual_dir" = "$WORKDIR" ] || fail "agent workdir mismatch: $actual_dir"
+
+MOSAIC_TMUX_SOCKET="$SOCKET" \
+MOSAIC_AGENT_WORKDIR="$WORKDIR" \
+MOSAIC_AGENT_COMMAND='bash --noprofile --norc -i' \
+  "$START" "$AGENT" >/tmp/mosaic-start-agent-idempotent.out
+
+grep -qF 'already running' /tmp/mosaic-start-agent-idempotent.out || fail "duplicate start was not idempotent"
+
+echo "ok - start-agent-session"

packages/mosaic/framework/tools/git/detect-platform.sh

@@ -169,6 +169,43 @@ raise SystemExit(1)
 PY
 }
 
+# Emit an actionable diagnostic to stderr when no tea login resolves for a host.
+# Callers that have a working API fallback may ignore the non-zero return of
+# get_gitea_login_for_host; this turns the previously SILENT failure into a loud,
+# greppable hint (available logins + override + add-login instructions). Printed to
+# stderr only, so it never contaminates stdout (the resolved login name) or log
+# assertions that capture tea/curl invocations.
+print_gitea_login_diagnostic() {
+    local host="${1:-<unknown>}"
+    local available
+    available=$(
+        command -v tea >/dev/null 2>&1 || { echo "(tea CLI not installed)"; exit 0; }
+        logins_json=$(tea login list --output json 2>/dev/null) || { echo "(could not query tea login list)"; exit 0; }
+        TEA_LOGINS_JSON="$logins_json" python3 - <<'PY'
+import json, os
+from urllib.parse import urlparse
+try:
+    logins = json.loads(os.environ.get("TEA_LOGINS_JSON", "[]"))
+except Exception:
+    logins = []
+rows = []
+for login in logins if isinstance(logins, list) else []:
+    name = str(login.get("name") or login.get("Name") or "")
+    url = str(login.get("url") or login.get("URL") or "")
+    host = urlparse(url).hostname or "?"
+    if name:
+        rows.append(f"{name} (host: {host})")
+print("; ".join(rows) if rows else "(none configured)")
+PY
+    )
+    {
+        echo "Error: no Gitea tea login matches host '$host'."
+        echo "  Available tea logins: ${available}"
+        echo "  Fix: set GITEA_LOGIN to a login whose URL host is '$host',"
+        echo "       or add one: tea login add --name <name> --url https://$host --token <token>"
+    } >&2
+}
+
 get_gitea_login_for_host() {
     local host="${1:-}"
     local login
@@ -190,6 +227,7 @@ get_gitea_login_for_host() {
         return 0
     fi
 
+    print_gitea_login_diagnostic "$host"
     return 1
 }
 

packages/mosaic/framework/tools/git/issue-assign.sh

@@ -98,27 +98,32 @@ case "$PLATFORM" in
         ;;
     gitea)
         # tea issue edit syntax
-        REPO_ARGS=$(get_gitea_repo_args) || {
-            echo "Error: Could not resolve Gitea repo/login args for remote host" >&2
+        REPO_SLUG=$(get_repo_slug) || {
+            echo "Error: Could not resolve Gitea repo slug from remote" >&2
             exit 1
         }
-        CMD="tea issue edit $ISSUE $REPO_ARGS"
+        REPO_LOGIN=$(get_gitea_login) || {
+            echo "Error: Could not resolve Gitea login for remote host" >&2
+            exit 1
+        }
+        REPO_ARGS=(--repo "$REPO_SLUG" --login "$REPO_LOGIN")
+        CMD=(tea issue edit "$ISSUE" "${REPO_ARGS[@]}")
         NEEDS_EDIT=false
 
         if [[ -n "$ASSIGNEE" ]]; then
             # tea uses --assignees flag
-            CMD="$CMD --assignees \"$ASSIGNEE\""
+            CMD+=(--assignees "$ASSIGNEE")
             NEEDS_EDIT=true
         fi
         if [[ -n "$LABELS" ]]; then
             # tea uses --labels flag (replaces existing)
-            CMD="$CMD --labels \"$LABELS\""
+            CMD+=(--labels "$LABELS")
             NEEDS_EDIT=true
         fi
         if [[ -n "$MILESTONE" ]]; then
-            MILESTONE_ID=$(tea milestones list $REPO_ARGS 2>/dev/null | grep -E "^\s*[0-9]+" | grep "$MILESTONE" | awk '{print $1}' | head -1)
+            MILESTONE_ID=$(tea milestones list "${REPO_ARGS[@]}" 2>/dev/null | grep -E "^\s*[0-9]+" | grep "$MILESTONE" | awk '{print $1}' | head -1)
             if [[ -n "$MILESTONE_ID" ]]; then
-                CMD="$CMD --milestone $MILESTONE_ID"
+                CMD+=(--milestone "$MILESTONE_ID")
                 NEEDS_EDIT=true
             else
                 echo "Warning: Could not find milestone '$MILESTONE'" >&2
@@ -126,7 +131,7 @@ case "$PLATFORM" in
         fi
 
         if [[ "$NEEDS_EDIT" == true ]]; then
-            eval "$CMD"
+            "${CMD[@]}"
             echo "Issue #$ISSUE updated successfully"
         else
             echo "No changes specified"

packages/mosaic/framework/tools/git/issue-comment.sh

@@ -53,7 +53,15 @@ if [[ "$PLATFORM" == "github" ]]; then
     gh issue comment "$ISSUE_NUMBER" --body "$COMMENT"
     echo "Added comment to GitHub issue #$ISSUE_NUMBER"
 elif [[ "$PLATFORM" == "gitea" ]]; then
-    tea issue comment "$ISSUE_NUMBER" "$COMMENT" $(get_gitea_repo_args)
+    # Build the invocation as an argv array (not unquoted $(get_gitea_repo_args)
+    # word-splitting) so the comment body — including Markdown backticks, $(...),
+    # and quotes — is passed verbatim and never re-split or shell-evaluated.
+    REPO_SLUG=$(get_repo_slug)
+    GITEA_LOGIN_NAME=$(get_gitea_login) || {
+        echo "Error: could not resolve a Gitea login for this repo; cannot comment on issue #$ISSUE_NUMBER." >&2
+        exit 1
+    }
+    tea issue comment "$ISSUE_NUMBER" "$COMMENT" --repo "$REPO_SLUG" --login "$GITEA_LOGIN_NAME"
     echo "Added comment to Gitea issue #$ISSUE_NUMBER"
 else
     echo "Error: Unknown platform"

packages/mosaic/framework/tools/git/issue-edit.sh

@@ -63,24 +63,28 @@ fi
 detect_platform >/dev/null
 
 if [[ "$PLATFORM" == "github" ]]; then
-    CMD="gh issue edit $ISSUE_NUMBER"
-    [[ -n "$TITLE" ]] && CMD="$CMD --title \"$TITLE\""
-    [[ -n "$BODY" ]] && CMD="$CMD --body \"$BODY\""
-    [[ -n "$LABELS" ]] && CMD="$CMD --add-label \"$LABELS\""
-    [[ -n "$MILESTONE" ]] && CMD="$CMD --milestone \"$MILESTONE\""
-    eval $CMD
+    CMD=(gh issue edit "$ISSUE_NUMBER")
+    [[ -n "$TITLE" ]] && CMD+=(--title "$TITLE")
+    [[ -n "$BODY" ]] && CMD+=(--body "$BODY")
+    [[ -n "$LABELS" ]] && CMD+=(--add-label "$LABELS")
+    [[ -n "$MILESTONE" ]] && CMD+=(--milestone "$MILESTONE")
+    "${CMD[@]}"
     echo "Updated GitHub issue #$ISSUE_NUMBER"
 elif [[ "$PLATFORM" == "gitea" ]]; then
-    REPO_ARGS=$(get_gitea_repo_args) || {
-        echo "Error: Could not resolve Gitea repo/login args for remote host" >&2
+    REPO_SLUG=$(get_repo_slug) || {
+        echo "Error: Could not resolve Gitea repo slug from remote" >&2
         exit 1
     }
-    CMD="tea issue edit $ISSUE_NUMBER $REPO_ARGS"
-    [[ -n "$TITLE" ]] && CMD="$CMD --title \"$TITLE\""
-    [[ -n "$BODY" ]] && CMD="$CMD --description \"$BODY\""
-    [[ -n "$LABELS" ]] && CMD="$CMD --add-labels \"$LABELS\""
-    [[ -n "$MILESTONE" ]] && CMD="$CMD --milestone \"$MILESTONE\""
-    eval $CMD
+    REPO_LOGIN=$(get_gitea_login) || {
+        echo "Error: Could not resolve Gitea login for remote host" >&2
+        exit 1
+    }
+    CMD=(tea issue edit "$ISSUE_NUMBER" --repo "$REPO_SLUG" --login "$REPO_LOGIN")
+    [[ -n "$TITLE" ]] && CMD+=(--title "$TITLE")
+    [[ -n "$BODY" ]] && CMD+=(--description "$BODY")
+    [[ -n "$LABELS" ]] && CMD+=(--add-labels "$LABELS")
+    [[ -n "$MILESTONE" ]] && CMD+=(--milestone "$MILESTONE")
+    "${CMD[@]}"
     echo "Updated Gitea issue #$ISSUE_NUMBER"
 else
     echo "Error: Unknown platform"

packages/mosaic/framework/tools/git/lane-brief.sh

@@ -0,0 +1,129 @@
+#!/usr/bin/env bash
+#
+# lane-brief.sh — live dispatch brief for a repo "lane" (milestone/label), straight
+# from current Gitea state. Defeats stale worker self-report: workers brief from
+# static notes and routinely report issues "todo" that are already CLOSED, forcing
+# the orchestrator to re-verify each one before dispatch. This returns the CURRENT
+# open set, classified for dispatch, in one call.
+#
+# Usage:
+#   lane-brief.sh -r <owner/repo> [-m <milestone>] [-l <label>] [-L <login>] [-n <limit>]
+#   lane-brief.sh -r usc/uconnect -m "M2M Part Search (0.0.45)"
+#   lane-brief.sh -r usc/uconnect -l domain/6-security
+#
+# Reliable signals (closed issues are excluded by definition — that's the point):
+#   - open-vs-closed : authoritative; this is the stale-intake failure mode.
+#   - PR-linkage     : an open PR referencing the issue = work underway.
+# Assignees/dependencies are intentionally NOT trusted as "available" signals —
+# fleets that track work-state out-of-band (tmux board, issue text) leave them
+# empty in Gitea. Output therefore partitions by PR presence and the OPEN-NO-PR set
+# is "dispatch candidates to cross-check against the live fleet", not a blind list.
+#
+# Login resolution order: -L flag > $GITEA_LOGIN > owner inference (usc->usc,
+# mosaicstack/mosaic->mosaicstack) > detect-platform.sh default-login fallback.
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+# shellcheck source=/dev/null
+source "$SCRIPT_DIR/detect-platform.sh"
+
+REPO="" MILESTONE="" LABEL="" LOGIN="" LIMIT=100
+while getopts "r:m:l:L:n:h" opt; do
+  case "$opt" in
+    r) REPO="$OPTARG" ;;
+    m) MILESTONE="$OPTARG" ;;
+    l) LABEL="$OPTARG" ;;
+    L) LOGIN="$OPTARG" ;;
+    n) LIMIT="$OPTARG" ;;
+    h) grep '^#' "$0" | sed 's/^# \?//'; exit 0 ;;
+    *) echo "see -h" >&2; exit 2 ;;
+  esac
+done
+[[ -n "$REPO" ]] || { echo "FATAL: -r <owner/repo> required" >&2; exit 2; }
+
+# Resolve login: explicit -L, then $GITEA_LOGIN, then owner inference, then the
+# shared default-login resolver. Owner inference comes before the shared fallback
+# because the latter is not owner-aware (picks the default tea login), which is
+# wrong for cross-instance lanes.
+if [[ -z "$LOGIN" ]]; then
+  if [[ -n "${GITEA_LOGIN:-}" ]]; then
+    LOGIN="$GITEA_LOGIN"
+  else
+    case "${REPO%%/*}" in
+      usc|USC) LOGIN=usc ;;
+      mosaicstack|mosaic) LOGIN=mosaicstack ;;
+      *) LOGIN="$(get_gitea_login_for_repo_override 2>/dev/null || true)" ;;
+    esac
+  fi
+fi
+[[ -n "$LOGIN" ]] || { echo "FATAL: could not resolve a Gitea login for $REPO (pass -L or set GITEA_LOGIN)" >&2; exit 2; }
+
+command -v tea >/dev/null || { echo "FATAL: tea not found" >&2; exit 1; }
+command -v jq  >/dev/null || { echo "FATAL: jq not found" >&2; exit 1; }
+
+ISSUES_JSON="$(tea issues list --repo "$REPO" --login "$LOGIN" --state open --limit "$LIMIT" \
+  --fields index,title,assignees,milestone,labels --output json 2>/dev/null)" || {
+  echo "FATAL: tea issues list failed for $REPO (login=$LOGIN)" >&2; exit 1; }
+
+# Open PRs, to cross-ref which issues already have work in flight. An issue is
+# "work underway" if an open PR links to it. Two link signals are honored:
+#   (a) a closing keyword in the PR BODY — Gitea's auto-close set (close/closes/
+#       closed, fix/fixes/fixed, resolve/resolves/resolved), case-insensitive,
+#       directly preceding `#N`. This is the AUTHORITATIVE link Gitea itself uses
+#       to associate a PR with the issue it resolves; a body-only "Closes #546"
+#       is the common case and MUST count. The earlier version inspected only the
+#       PR index/title/head TSV (never the body or Gitea linkage), so a body-only
+#       reference was invisible and the linked OPEN issue was misclassified as a
+#       dispatch candidate — re-dispatchable in-flight work (the #546/#547 defect).
+#   (b) a bare #N in the PR title, or an issue number embedded in the head branch
+#       (feat/546-x, fix-546) — the weaker heuristic preserved from prior behavior.
+# Bare #N mentions in the BODY are deliberately NOT treated as links: PR bodies
+# routinely name unrelated issues in prose ("relevant to the #538 line of work"),
+# and counting those would wrongly mark live, dispatchable issues as in-flight.
+# Only the closing-keyword form is a commitment to resolve that issue. Requiring
+# `#` to directly follow the keyword also keeps cross-repo `owner/repo#N` forms
+# from leaking a foreign issue number into this per-repo lane (cross-repo lanes
+# are run per-repo). JSON (not TSV) is used so multi-line bodies parse cleanly.
+PRS_JSON="$(tea pulls list --repo "$REPO" --login "$LOGIN" --state open \
+  --fields index,title,head,body --output json 2>/dev/null || echo '[]')"
+[[ -n "$PRS_JSON" ]] || PRS_JSON='[]'
+
+# \b anchors the keyword to a word start so embedded substrings do not match
+# (e.g. "prefix #5", "disclosed #7" must NOT be read as "fix #5" / "closed #7").
+GITEA_CLOSE_KW='close[sd]?|fix(e[sd])?|resolve[sd]?'
+PR_BODY_REFS="$(printf '%s' "$PRS_JSON" | jq -r '.[] | .body // ""' 2>/dev/null \
+  | grep -oiE "\\b(${GITEA_CLOSE_KW})[[:space:]:]+#[0-9]+" | grep -oE '[0-9]+' || true)"
+PR_TITLE_HEAD_REFS="$(printf '%s' "$PRS_JSON" \
+  | jq -r '.[] | [ (.title // ""), (.head // "" | if type=="object" then (.ref // "") else . end) ] | join(" ")' 2>/dev/null \
+  | grep -oE '#[0-9]+|[/-][0-9]{3,}' | grep -oE '[0-9]+' || true)"
+PR_ISSUE_REFS="$(printf '%s\n%s\n' "$PR_BODY_REFS" "$PR_TITLE_HEAD_REFS" | grep -E '^[0-9]+$' | sort -u || true)"
+
+ts="$(date -u '+%Y-%m-%d %H:%MZ' 2>/dev/null || echo '?')"
+filt="$REPO"; [[ -n "$MILESTONE" ]] && filt="$filt · milestone:'$MILESTONE'"; [[ -n "$LABEL" ]] && filt="$filt · label:'$LABEL'"
+echo "LANE BRIEF — $filt · $ts (login=$LOGIN)"
+echo "(open issues only; closed are excluded by definition — that's the point)"
+echo
+
+# Label match is exact-token against tea's space-separated labels string (so -l
+# "security" does NOT match label "domain/6-security"). Caveat: label names that
+# themselves contain spaces aren't distinguishable in tea's string form.
+printf '%s' "$ISSUES_JSON" | jq -r --arg ms "$MILESTONE" --arg lb "$LABEL" --arg prs "$PR_ISSUE_REFS" '
+  ($prs | split("\n") | map(select(length>0))) as $prrefs
+  | map(
+      select( ($ms=="" or .milestone==$ms)
+          and ($lb=="" or ((.labels//"") | split(" ") | index($lb) != null)) )
+      | . + { assigned: ((.assignees//"")|length>0),
+              haspr: (.index as $ix | ($prrefs | index($ix)) != null) }
+    )
+  | (map(select(.haspr|not)))  as $candidates
+  | (map(select(.haspr)))      as $inflight
+  | "DISPATCH CANDIDATES (open · no open PR) — \($candidates|length)  [cross-check vs live fleet]:",
+    ( $candidates[] | "  #\(.index)  \(.title[0:90])\(if .assigned then "  (gitea-assignee set)" else "" end)" ),
+    "",
+    "WORK UNDERWAY (open · PR in flight) — \($inflight|length):",
+    ( $inflight[] | "  #\(.index)  \(.title[0:80])  [PR open]" )
+'
+echo
+echo "Closed issues are excluded — do NOT take a worker's self-reported 'todo' on faith."
+echo "Candidates = open + no PR; confirm against the live fleet before dispatch"
+echo "(fleets that don't self-assign in Gitea leave 'unassigned' meaningless)."

packages/mosaic/framework/tools/git/milestone-create.sh

@@ -99,10 +99,15 @@ fi
 case "$PLATFORM" in
     github)
         # GitHub uses the API for milestone creation
-        JSON_PAYLOAD="{\"title\":\"$TITLE\""
-        [[ -n "$DESCRIPTION" ]] && JSON_PAYLOAD="$JSON_PAYLOAD,\"description\":\"$DESCRIPTION\""
-        [[ -n "$DUE_DATE" ]] && JSON_PAYLOAD="$JSON_PAYLOAD,\"due_on\":\"${DUE_DATE}T00:00:00Z\""
-        JSON_PAYLOAD="$JSON_PAYLOAD}"
+        # Use jq to safely construct JSON so titles/descriptions containing
+        # quotes or special characters do not corrupt the payload (F-07).
+        JSON_PAYLOAD=$(jq -n \
+            --arg t "$TITLE" \
+            --arg d "$DESCRIPTION" \
+            --arg due "${DUE_DATE}" \
+            '{"title": $t}
+             + (if $d != "" then {"description": $d} else {} end)
+             + (if $due != "" then {"due_on": ($due + "T00:00:00Z")} else {} end)')
 
         gh api repos/:owner/:repo/milestones --method POST --input - <<< "$JSON_PAYLOAD"
         echo "Milestone '$TITLE' created successfully"

packages/mosaic/framework/tools/git/pr-ci-wait.sh

@@ -72,6 +72,11 @@ elif values and all(v == "success" for v in values):
     print("success")
 elif any(v in {"pending", "running", "queued", "waiting"} for v in values):
     print("pending")
+elif not values and not state:
+    # No pipeline/status of any kind reported for this commit. Distinct from
+    # "unknown" (an ambiguous/unrecognized status that should keep polling):
+    # this signals a repo/commit that simply has no CI configured.
+    print("no-status")
 else:
     print("unknown")
 PY
@@ -142,6 +147,21 @@ gitea_get_commit_status_json() {
     curl -fsSL -H "User-Agent: curl/8" -H "Authorization: token ${token}" "$url"
 }
 
+gitea_get_default_branch() {
+    local host="$1"
+    local repo="$2"
+    local token="$3"
+    local url="https://${host}/api/v1/repos/${repo}"
+    curl -fsSL -H "User-Agent: curl/8" -H "Authorization: token ${token}" "$url" | python3 -c '
+import json, sys
+print((json.load(sys.stdin) or {}).get("default_branch", ""))
+'
+}
+
+github_get_default_branch() {
+    gh api "repos/${OWNER}/${REPO}" --jq '.default_branch'
+}
+
 while [[ $# -gt 0 ]]; do
     case "$1" in
         -n|--number)
@@ -245,6 +265,51 @@ else
     exit 1
 fi
 
+# No-CI determination is TWO-TIER (primary: CI history; secondary: empty-poll streak).
+#
+# PRIMARY — "does this repo run CI at all?" Probed once, up front, from the DEFAULT
+# BRANCH's commit status. A repo whose default branch carries CI statuses
+# demonstrably runs CI, so an EMPTY status on the PR head means the pipeline simply
+# has not registered YET (webhook/queue lag) — NOT that the repo is CI-less. In that
+# case we must NEVER fast-green; we keep polling until the pipeline registers or the
+# timeout fires (both safe). This closes the webhook-lag false-green: a slow-to-
+# register pipeline feeding a merge gate can no longer be mistaken for "no CI".
+#
+# SECONDARY — the empty-poll streak below applies ONLY to genuinely CI-less repos
+# (default branch also has no CI history, e.g. device-imaging class), where burning
+# the full timeout would be pure waste. There, NO_CI_MAX empty polls => fast-exit 0.
+#
+# Probe failure is treated conservatively as REPO_HAS_CI=1 (assume CI present): we
+# would rather wait-then-timeout than risk a false-green, per the merge-gate priority.
+REPO_HAS_CI=1
+detect_repo_ci() {
+    local def_branch def_status
+    # Every early exit returns 0: a probe miss must leave the conservative
+    # REPO_HAS_CI=1 default in place, never abort the caller under `set -e`.
+    if [[ "$PLATFORM" == "github" ]]; then
+        def_branch=$(github_get_default_branch 2>/dev/null) || {
+            echo "[pr-ci-wait] WARN: default-branch probe failed; assuming CI-enabled (will not fast-green on empty status)."; return 0; }
+        [[ -n "$def_branch" ]] || return 0
+        def_status=$(github_get_commit_status_json "$OWNER" "$REPO" "$def_branch" 2>/dev/null | extract_state_from_status_json) || return 0
+    else
+        def_branch=$(gitea_get_default_branch "$HOST" "$OWNER/$REPO" "$TOKEN" 2>/dev/null) || {
+            echo "[pr-ci-wait] WARN: default-branch probe failed; assuming CI-enabled (will not fast-green on empty status)."; return 0; }
+        [[ -n "$def_branch" ]] || return 0
+        def_status=$(gitea_get_commit_status_json "$HOST" "$OWNER/$REPO" "$TOKEN" "$def_branch" 2>/dev/null | extract_state_from_status_json) || return 0
+    fi
+    if [[ "$def_status" == "no-status" || -z "$def_status" ]]; then
+        REPO_HAS_CI=0
+        echo "[pr-ci-wait] default branch '${def_branch}' has no CI status history — treating repo as CI-less (empty-poll fast-exit enabled)."
+    else
+        REPO_HAS_CI=1
+        echo "[pr-ci-wait] default branch '${def_branch}' has CI history (state=${def_status}) — repo runs CI; empty status on PR head => awaiting registration, will not fast-green."
+    fi
+}
+detect_repo_ci || true
+
+NO_CI_STREAK=0
+NO_CI_MAX=3
+
 while true; do
     NOW_TS=$(date +%s)
     if (( NOW_TS > DEADLINE_TS )); then
@@ -272,11 +337,35 @@ while true; do
             echo "Error: CI reported ${STATE} for PR #$PR_NUMBER." >&2
             exit 1
             ;;
+        no-status)
+            if [[ "$REPO_HAS_CI" == "1" ]]; then
+                # PRIMARY tier: repo demonstrably runs CI but this commit's pipeline
+                # has not registered yet (webhook/queue lag). Do NOT fast-green — keep
+                # polling until it registers or the timeout fires. Reset the streak so
+                # a later genuine CI-less misread can't accumulate across this state.
+                NO_CI_STREAK=0
+                echo "[pr-ci-wait] empty status on PR head but repo runs CI — awaiting pipeline registration (webhook lag), not fast-greening."
+            else
+                # SECONDARY tier: genuinely CI-less repo (default branch has no CI
+                # history either). Empty polls => fast-exit green after NO_CI_MAX.
+                NO_CI_STREAK=$((NO_CI_STREAK + 1))
+                if (( NO_CI_STREAK >= NO_CI_MAX )); then
+                    echo "[INFO] no CI configured for this repo/commit (PR #$PR_NUMBER, ${NO_CI_STREAK} consecutive empty polls, default branch also CI-less); treating as green."
+                    exit 0
+                fi
+            fi
+            sleep "$INTERVAL_SEC"
+            ;;
         pending|unknown)
+            # A pipeline exists but hasn't reached a terminal state (or is
+            # transiently ambiguous) — keep waiting, and reset the no-CI streak
+            # since this commit is not in the "no CI at all" condition.
+            NO_CI_STREAK=0
             sleep "$INTERVAL_SEC"
             ;;
         *)
             echo "[pr-ci-wait] Unrecognized state '${STATE}', continuing to poll..."
+            NO_CI_STREAK=0
             sleep "$INTERVAL_SEC"
             ;;
     esac

packages/mosaic/framework/tools/git/pr-metadata.sh

@@ -57,12 +57,20 @@ curl_gitea_pull() {
     local token basic_auth raw_code body_file http_code
     body_file=$(mktemp)
 
+    # shellcheck disable=SC2329 # Invoked by the RETURN trap below.
+    cleanup_gitea_pull_body() {
+        local status=$?
+        rm -f -- "$body_file"
+        trap - RETURN
+        return "$status"
+    }
+    trap cleanup_gitea_pull_body RETURN
+
     token=$(get_gitea_token "$HOST" || true)
     if [[ -n "$token" ]]; then
         raw_code=$(curl -sS -w '%{http_code}' -o "$body_file" -H "User-Agent: curl/8" -H "Authorization: token $token" "$api_url" || true)
         if [[ "$raw_code" =~ ^2 ]]; then
-            cat "$body_file"
-            rm -f "$body_file"
+            cat "$body_file" || return $?
             return 0
         fi
         http_code="$raw_code"
@@ -72,8 +80,7 @@ curl_gitea_pull() {
     if [[ -n "$basic_auth" ]]; then
         raw_code=$(curl -sS -w '%{http_code}' -o "$body_file" -u "$basic_auth" -H "User-Agent: curl/8" "$api_url" || true)
         if [[ "$raw_code" =~ ^2 ]]; then
-            cat "$body_file"
-            rm -f "$body_file"
+            cat "$body_file" || return $?
             return 0
         fi
         http_code="$raw_code"
@@ -96,7 +103,6 @@ except Exception:
     message = open(path, encoding="utf-8", errors="replace").read()[:200] or "empty response"
 print(f"Error: Gitea pull request API request failed with HTTP {code}: {message}")
 PY
-    rm -f "$body_file"
     return 1
 }
 

packages/mosaic/framework/tools/git/test-gitea-login-resolution.sh

@@ -230,4 +230,81 @@ if grep -q -- 'tea issue close 536 .*--login mosaicstack' "$LOG_FILE"; then
     exit 1
 fi
 
+# ---------------------------------------------------------------------------
+# #560: loud diagnostic + host-derived login for BOTH instances + override-wins
+# ---------------------------------------------------------------------------
+
+# Loud diagnostic: a host with no matching tea login must emit an actionable
+# error to stderr (the previous behavior was a SILENT failure). The original
+# mock defines only usc/evil-usc logins, so mosaicstack resolution fails here.
+git -C "$REPO_DIR" remote set-url origin https://git.mosaicstack.dev/mosaicstack/stack.git
+diag_stderr=$(run_in_repo bash -c '
+    source "'"$SCRIPT_DIR"'/detect-platform.sh"
+    get_gitea_login_for_host git.mosaicstack.dev
+' 2>&1 1>/dev/null || true)
+if ! grep -q "no Gitea tea login matches host 'git.mosaicstack.dev'" <<<"$diag_stderr"; then
+    echo "Expected loud diagnostic naming the unresolved host; got: $diag_stderr" >&2
+    exit 1
+fi
+if ! grep -q "Available tea logins:" <<<"$diag_stderr"; then
+    echo "Expected diagnostic to list available tea logins; got: $diag_stderr" >&2
+    exit 1
+fi
+
+# Both-instance host derivation + override-wins, using a mock that DOES define a
+# mosaicstack login. Scoped to this section so the API-fallback assertions above
+# (which rely on mosaicstack having NO tea login) remain valid.
+BIN_DIR2="$WORK_DIR/bin2"
+mkdir -p "$BIN_DIR2"
+cp "$BIN_DIR/curl" "$BIN_DIR2/curl"
+cat > "$BIN_DIR2/tea" <<'SH'
+#!/usr/bin/env bash
+set -euo pipefail
+if [[ "$*" == "login list --output json" ]]; then
+    cat <<'JSON'
+[
+  {"name":"mosaicstack","url":"https://git.mosaicstack.dev","user":"jason.woltje"},
+  {"name":"usc","url":"https://git.uscllc.com","user":"jason.woltje"}
+]
+JSON
+    exit 0
+fi
+printf 'tea %s\n' "$*" >> "$MOSAIC_TEST_LOG"
+exit 0
+SH
+chmod +x "$BIN_DIR2/tea"
+
+run_in_repo2() {
+    (
+        cd "$REPO_DIR"
+        PATH="$BIN_DIR2:$PATH" \
+        MOSAIC_CREDENTIALS_FILE="$CREDENTIALS_FILE" \
+        MOSAIC_TEST_LOG="$LOG_FILE" \
+        "$@"
+    )
+}
+
+git -C "$REPO_DIR" remote set-url origin https://git.mosaicstack.dev/mosaicstack/stack.git
+mosaic_login=$(run_in_repo2 bash -c 'source "'"$SCRIPT_DIR"'/detect-platform.sh"; get_gitea_login')
+if [[ "$mosaic_login" != "mosaicstack" ]]; then
+    echo "Expected mosaicstack origin to derive login 'mosaicstack'; got '$mosaic_login'" >&2
+    exit 1
+fi
+
+git -C "$REPO_DIR" remote set-url origin https://git.uscllc.com/USC/uconnect.git
+usc_login_derived=$(run_in_repo2 bash -c 'source "'"$SCRIPT_DIR"'/detect-platform.sh"; get_gitea_login')
+if [[ "$usc_login_derived" != "usc" ]]; then
+    echo "Expected usc origin to derive login 'usc'; got '$usc_login_derived'" >&2
+    exit 1
+fi
+
+# Explicit GITEA_LOGIN override is honored when it matches the host.
+git -C "$REPO_DIR" remote set-url origin https://git.mosaicstack.dev/mosaicstack/stack.git
+override_wins=$(run_in_repo2 bash -c 'export GITEA_LOGIN=mosaicstack; source "'"$SCRIPT_DIR"'/detect-platform.sh"; get_gitea_login')
+if [[ "$override_wins" != "mosaicstack" ]]; then
+    echo "Expected valid GITEA_LOGIN override to win on mosaicstack host; got '$override_wins'" >&2
+    exit 1
+fi
+git -C "$REPO_DIR" remote set-url origin https://git.uscllc.com/USC/uconnect.git
+
 echo "Gitea login resolution regression harness passed"

packages/mosaic/framework/tools/git/test-issue-create-body-safety.sh

@@ -0,0 +1,102 @@
+#!/usr/bin/env bash
+# Regression harness for issue-create.sh Markdown-body safety (#559).
+#
+# Guards against reintroduction of eval-based command construction. The wrapper
+# builds its tea/gh invocation as an argv array, so a body containing command
+# substitution ($(...)), backticks, quotes, and dollar signs MUST reach tea
+# verbatim and MUST NOT be shell-evaluated. This test asserts both:
+#   1. No command-substitution side effect (an injected `touch SENTINEL` never runs).
+#   2. The --description value tea receives is byte-for-byte the original body.
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/issue-create-body-safety}"
+REPO_DIR="$WORK_DIR/repo"
+BIN_DIR="$WORK_DIR/bin"
+SENTINEL="$WORK_DIR/INJECTION_SENTINEL"
+BODY_FILE="$WORK_DIR/body.txt"
+RECEIVED_FILE="$WORK_DIR/received-description.txt"
+
+rm -rf "$WORK_DIR"
+mkdir -p "$REPO_DIR" "$BIN_DIR"
+
+git -C "$REPO_DIR" init -q
+git -C "$REPO_DIR" remote add origin https://git.mosaicstack.dev/mosaicstack/stack.git
+
+# Hostile Markdown body. The unquoted heredoc expands $SENTINEL (a real path we
+# want embedded) but every shell metacharacter we care about is backslash-escaped
+# so the TEST shell writes them literally into the file — the bytes the wrapper
+# must then preserve.
+cat > "$BODY_FILE" <<EOF
+# Release notes
+
+Inline code: \`rm -rf /\` must stay literal.
+Command sub attempt: \$(touch $SENTINEL)
+Backtick cmd attempt: \`touch $SENTINEL\`
+Dollars: \$HOME \${PATH} \$5.00 and 100% done
+Quotes: "double" and 'single' and \`mixed\`
+Trailing pipe-ish: foo | bar && baz ; qux
+EOF
+
+BODY="$(cat "$BODY_FILE")"
+
+# Mock tea: resolve a mosaicstack login, then capture the --description verbatim.
+cat > "$BIN_DIR/tea" <<'SH'
+#!/usr/bin/env bash
+set -euo pipefail
+
+if [[ "$*" == "login list --output json" ]]; then
+    cat <<'JSON'
+[
+  {"name":"mosaicstack","url":"https://git.mosaicstack.dev","user":"jason.woltje"}
+]
+JSON
+    exit 0
+fi
+
+if [[ "${1:-}" == "issue" && "${2:-}" == "create" ]]; then
+    desc=""
+    while [[ $# -gt 0 ]]; do
+        case "$1" in
+            --description) desc="$2"; shift 2 ;;
+            *) shift ;;
+        esac
+    done
+    printf '%s' "$desc" > "$MOSAIC_TEST_RECEIVED"
+    echo "#1 created"
+    exit 0
+fi
+
+exit 0
+SH
+chmod +x "$BIN_DIR/tea"
+
+(
+    cd "$REPO_DIR"
+    PATH="$BIN_DIR:$PATH" \
+    MOSAIC_TEST_RECEIVED="$RECEIVED_FILE" \
+    "$SCRIPT_DIR/issue-create.sh" -t "Body safety test" -b "$BODY"
+) >/dev/null
+
+# 1. No command substitution executed anywhere in the pipeline.
+if [[ -e "$SENTINEL" ]]; then
+    echo "FAIL: injected command substitution executed (sentinel file created): $SENTINEL" >&2
+    exit 1
+fi
+
+# 2. tea actually received the body (issue create path taken, not silently dropped).
+if [[ ! -f "$RECEIVED_FILE" ]]; then
+    echo "FAIL: tea issue create was never invoked with a --description" >&2
+    exit 1
+fi
+
+# 3. The description tea received is byte-for-byte the original body.
+if [[ "$(cat "$RECEIVED_FILE")" != "$BODY" ]]; then
+    echo "FAIL: body was not preserved verbatim through issue-create.sh" >&2
+    echo "--- expected ---" >&2; printf '%s\n' "$BODY" >&2
+    echo "--- received ---" >&2; cat "$RECEIVED_FILE" >&2
+    exit 1
+fi
+
+echo "issue-create.sh Markdown body-safety regression harness passed"

packages/mosaic/framework/tools/git/test-lane-brief-pr-linkage.sh

@@ -0,0 +1,114 @@
+#!/usr/bin/env bash
+# Regression harness for lane-brief.sh PR->issue linkage classification.
+#
+# Covers the #546/#547 defect: lane-brief.sh inspected only the PR index/title/head
+# fields and never the PR BODY, so an open PR whose body says "Closes #546" did not
+# mark issue #546 as work-underway — #546 was listed as a DISPATCH CANDIDATE and was
+# re-dispatchable in-flight work.
+#
+# Asserts:
+#   1. an open issue closed-keyword-linked from a PR BODY ("Closes #546") is
+#      classified WORK UNDERWAY, not a dispatch candidate.
+#   2. a BARE "#777" prose mention in a PR body does NOT classify #777 as
+#      work-underway (only Gitea closing keywords are a real link) — #777 stays a
+#      dispatch candidate.
+#   3. NON-VACUITY / RED-ON-REVERT: a copy of the script with the body-scan removed
+#      misclassifies #546 as a dispatch candidate — proving the body-scan is exactly
+#      what fixes the defect and that assertion 1 fails if the fix is reverted.
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+LANE_BRIEF="$SCRIPT_DIR/lane-brief.sh"
+WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/lane-brief-pr-linkage}"
+BIN_DIR="$WORK_DIR/bin"
+
+rm -rf "$WORK_DIR"
+mkdir -p "$BIN_DIR"
+
+# --- fake `tea`: serves a fixed open-issue set and one open PR. ----------------
+# PR #547 body uses a closing keyword for #546 ("Closes #546") and a BARE mention
+# of #777 ("the #777 line of work"). #777 must NOT be treated as linked.
+cat > "$BIN_DIR/tea" <<'SH'
+#!/usr/bin/env bash
+set -euo pipefail
+case "${1:-} ${2:-}" in
+  "issues list")
+    cat <<'JSON'
+[
+  {"index":"546","title":"lane-brief + ci-wait orchestration tooling","assignees":[],"milestone":null,"labels":""},
+  {"index":"777","title":"unrelated downstream item","assignees":[],"milestone":null,"labels":""},
+  {"index":"999","title":"item only named inside the word hotfix","assignees":[],"milestone":null,"labels":""}
+]
+JSON
+    ;;
+  "pulls list")
+    cat <<'JSON'
+[
+  {"index":"547","title":"feat(framework/tools): orchestration helpers","head":"feat/orchestration-tools-lane-brief-ci-wait","body":"Two additive orchestration tools.\n\nCloses #546.\n\nLogin resolution is relevant to the #777 line of work but does not touch it.\nThis shipped as a hotfix #999 earlier — that bare reference must not link it.\n\nFixes #546\n"}
+]
+JSON
+    ;;
+  *)
+    echo "fake-tea: unhandled: $*" >&2; exit 1 ;;
+esac
+SH
+chmod +x "$BIN_DIR/tea"
+
+run_brief() { # $1 = script path
+  PATH="$BIN_DIR:$PATH" "$1" -r mosaic/stack -L test-login 2>/dev/null
+}
+
+# Extract the issue numbers under a named section header until the next blank line.
+section_nums() { # $1 = output  $2 = header-prefix
+  printf '%s\n' "$1" | awk -v h="$2" '
+    index($0,h)==1 {grab=1; next}
+    grab && /^[[:space:]]*$/ {grab=0}
+    grab && match($0, /#[0-9]+/) { print substr($0, RSTART+1, RLENGTH-1) }
+  '
+}
+
+fail() { echo "FAIL: $1" >&2; exit 1; }
+contains() { printf '%s\n' "$1" | grep -qx "$2"; }
+
+# ---------------------------------------------------------------------------
+# Fixed (current) script behavior
+# ---------------------------------------------------------------------------
+OUT="$(run_brief "$LANE_BRIEF")"
+CAND="$(section_nums "$OUT" 'DISPATCH CANDIDATES')"
+UNDER="$(section_nums "$OUT" 'WORK UNDERWAY')"
+
+echo "--- lane-brief output (fixed) ---"; printf '%s\n' "$OUT"
+echo "--- candidates: [$(printf '%s' "$CAND" | tr '\n' ' ')] underway: [$(printf '%s' "$UNDER" | tr '\n' ' ')] ---"
+
+contains "$UNDER" 546 || fail "#546 (PR body 'Closes #546') should be WORK UNDERWAY"
+contains "$CAND" 546 && fail "#546 must NOT be a dispatch candidate (it has an open PR)"
+contains "$CAND" 777 || fail "#777 (only a bare prose mention) should remain a dispatch candidate"
+contains "$UNDER" 777 && fail "#777 must NOT be work-underway — bare body mentions are not links"
+contains "$CAND" 999 || fail "#999 ('hotfix #999' — keyword is a substring) should remain a candidate"
+contains "$UNDER" 999 && fail "#999 must NOT be work-underway — word-boundary must reject 'hotfix'"
+echo "PASS: body closing-keyword link classifies #546 underway; bare #777 / substring #999 stay candidates"
+
+# ---------------------------------------------------------------------------
+# NON-VACUITY: revert the body-scan and prove #546 regresses to a candidate.
+# ---------------------------------------------------------------------------
+REVERTED="$SCRIPT_DIR/.lane-brief.reverted.$$.sh"
+trap 'rm -f "$REVERTED"' EXIT
+# Drop the PR_BODY_REFS contribution from the union (simulates the pre-fix script
+# that only looked at index/title/head). Sibling `source detect-platform.sh` still
+# resolves because the copy lives in the same dir.
+# shellcheck disable=SC2016  # single-quoted on purpose: sed needs the literal $PR_BODY_REFS
+sed 's/"\$PR_BODY_REFS"/""/' "$LANE_BRIEF" > "$REVERTED"
+chmod +x "$REVERTED"
+grep -q 'PR_BODY_REFS' "$REVERTED" || fail "revert sed anchor not found — test is stale"
+
+ROUT="$(run_brief "$REVERTED")"
+RCAND="$(section_nums "$ROUT" 'DISPATCH CANDIDATES')"
+RUNDER="$(section_nums "$ROUT" 'WORK UNDERWAY')"
+echo "--- candidates(reverted): [$(printf '%s' "$RCAND" | tr '\n' ' ')] underway: [$(printf '%s' "$RUNDER" | tr '\n' ' ')] ---"
+
+contains "$RCAND" 546 || fail "non-vacuity broken: reverted script should misclassify #546 as a candidate"
+contains "$RUNDER" 546 && fail "non-vacuity broken: reverted script should NOT mark #546 underway"
+echo "PASS (RED-on-revert): without the body-scan, #546 regresses to a dispatch candidate"
+
+echo "ALL PASS: test-lane-brief-pr-linkage.sh"

packages/mosaic/framework/tools/git/test-pr-metadata-gitea.sh

@@ -7,9 +7,10 @@ SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/pr-metadata-gitea}"
 REPO_DIR="$WORK_DIR/repo"
 FIXTURE_DIR="$WORK_DIR/fixtures"
+STUB_DIR="$WORK_DIR/stubs"
 
 rm -rf "$WORK_DIR"
-mkdir -p "$REPO_DIR" "$FIXTURE_DIR"
+mkdir -p "$REPO_DIR" "$FIXTURE_DIR" "$STUB_DIR"
 
 git -C "$REPO_DIR" init -q
 git -C "$REPO_DIR" remote add origin https://git.uscllc.com/USC/uconnect.git
@@ -56,6 +57,150 @@ cat > "$FIXTURE_DIR/gitea-error.json" <<'JSON'
 {"message": "user does not exist [uid: 0, name: ]", "url": "https://git.uscllc.com/api/swagger"}
 JSON
 
+cat > "$STUB_DIR/curl" <<'SH'
+#!/usr/bin/env bash
+set -euo pipefail
+
+output_file=""
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        -o)
+            output_file="$2"
+            shift 2
+            ;;
+        -w|-H|-u)
+            shift 2
+            ;;
+        -s|-S|-sS)
+            shift
+            ;;
+        *)
+            shift
+            ;;
+    esac
+done
+
+if [[ -z "$output_file" ]]; then
+    echo "curl stub expected -o <output_file>" >&2
+    exit 2
+fi
+
+case "${MOSAIC_STUB_CURL_MODE:-success}" in
+    success)
+        cat > "$output_file" <<'JSON'
+{
+  "number": 1910,
+  "title": "Live curl path",
+  "state": "open",
+  "user": {"login": "edith"},
+  "head": {"ref": "fix/live-curl-path"},
+  "base": {"ref": "main"},
+  "html_url": "https://git.example.test/acme/widgets/pulls/1910"
+}
+JSON
+        printf '200'
+        ;;
+    cat-fails-after-2xx)
+        rm -f -- "$output_file"
+        ln -s /nonexistent/pr-metadata-body "$output_file"
+        printf '200'
+        ;;
+    *)
+        echo "unknown MOSAIC_STUB_CURL_MODE=${MOSAIC_STUB_CURL_MODE:-}" >&2
+        exit 2
+        ;;
+esac
+SH
+chmod +x "$STUB_DIR/curl"
+
+assert_tmpdir_empty() {
+    local tmpdir="$1" leftover
+    leftover=$(find "$tmpdir" -mindepth 1 -print -quit)
+    if [[ -n "$leftover" ]]; then
+        echo "Expected tmpfile cleanup, found leftover: $leftover" >&2
+        find "$tmpdir" -mindepth 1 -maxdepth 1 -ls >&2
+        exit 1
+    fi
+}
+
+run_curl_success_case() {
+    local tmpdir="$WORK_DIR/tmp-success" stderr_file="$WORK_DIR/curl-success.stderr"
+    local output status
+    mkdir -p "$tmpdir"
+
+    set +e
+    output=$(cd "$REPO_DIR" && \
+        PATH="$STUB_DIR:$PATH" \
+        TMPDIR="$tmpdir" \
+        GITEA_TOKEN="stub-token" \
+        GITEA_URL="https://git.example.test" \
+        MOSAIC_STUB_CURL_MODE="success" \
+        "$SCRIPT_DIR/pr-metadata.sh" -n 1910 2>"$stderr_file")
+    status=$?
+    set -e
+
+    if [[ "$status" -ne 0 ]]; then
+        echo "Expected curl success path to pass, got status $status" >&2
+        cat "$stderr_file" >&2
+        exit 1
+    fi
+    if grep -q "unbound variable" "$stderr_file"; then
+        echo "curl success path emitted unbound-variable cleanup noise" >&2
+        cat "$stderr_file" >&2
+        exit 1
+    fi
+    assert_tmpdir_empty "$tmpdir"
+
+    PR_METADATA_OUTPUT="$output" python3 - <<'PY'
+import json
+import os
+
+data = json.loads(os.environ["PR_METADATA_OUTPUT"])
+assert data["number"] == 1910, data
+assert data["baseRefName"] == "main", data
+assert data["headRefName"] == "fix/live-curl-path", data
+PY
+}
+
+run_curl_early_exit_cleanup_case() {
+    local tmpdir="$WORK_DIR/tmp-early-exit" stderr_file="$WORK_DIR/curl-early-exit.stderr"
+    local output status
+    mkdir -p "$tmpdir"
+
+    set +e
+    output=$(cd "$REPO_DIR" && \
+        PATH="$STUB_DIR:$PATH" \
+        TMPDIR="$tmpdir" \
+        GITEA_TOKEN="stub-token" \
+        GITEA_URL="https://git.example.test" \
+        MOSAIC_STUB_CURL_MODE="cat-fails-after-2xx" \
+        "$SCRIPT_DIR/pr-metadata.sh" -n 1910 2>"$stderr_file")
+    status=$?
+    set -e
+
+    if [[ "$status" -eq 0 ]]; then
+        echo "Expected unreadable 2xx body path to fail" >&2
+        printf '%s\n' "$output" >&2
+        exit 1
+    fi
+    if grep -q "unbound variable" "$stderr_file"; then
+        echo "curl early-exit path emitted unbound-variable cleanup noise" >&2
+        cat "$stderr_file" >&2
+        exit 1
+    fi
+    if ! grep -q "No such file or directory" "$stderr_file"; then
+        echo "Expected body-read failure from broken symlink path" >&2
+        cat "$stderr_file" >&2
+        exit 1
+    fi
+    if grep -q "Gitea API returned non-JSON" "$stderr_file"; then
+        echo "curl helper masked body-read failure as later JSON parsing failure" >&2
+        cat "$stderr_file" >&2
+        exit 1
+    fi
+    assert_tmpdir_empty "$tmpdir"
+}
+
 run_case() {
     local fixture="$1" expected_number="$2" expected_head="$3"
     local output
@@ -77,6 +222,8 @@ PY
 run_case "$FIXTURE_DIR/gitea-standard.json" 1905 edith/t_39ce717c-authentik-smoke-gate
 run_case "$FIXTURE_DIR/gitea-fallback.json" 1908 fix/fallback-head
 run_case "$FIXTURE_DIR/gitea-refs-pull-label.json" 1908 fix/t_23fa9e1d-portal-health-backend
+run_curl_success_case
+run_curl_early_exit_cleanup_case
 
 if cd "$REPO_DIR" && MOSAIC_GITEA_PR_METADATA_RAW_FILE="$FIXTURE_DIR/gitea-error.json" "$SCRIPT_DIR/pr-metadata.sh" -n 1909 >/dev/null 2>"$WORK_DIR/error.log"; then
     echo "Expected API error fixture to fail" >&2

packages/mosaic/framework/tools/tmux/README.md

@@ -31,9 +31,12 @@ Prepends the preamble automatically (auto-detecting your own `host:session`) and
 delivers reliably to local OR remote panes.
 
 ```bash
-# Local target (same host)
+# Local target (same host, default tmux server)
 agent-send.sh -s <dst_session> -m "message"
 
+# Local target on a Mosaic fleet socket
+agent-send.sh -L mosaic-factory -s '=coder0' -m "message"
+
 # Remote target (over ssh)
 agent-send.sh -H user@host -s <dst_session> -m "message"
 
@@ -42,10 +45,27 @@ agent-send.sh -H user@host -s <dst_session> -f msg.txt
 echo "msg" | agent-send.sh -s <dst_session>
 ```
 
-Key flags: `-s` dst session (required) · `-H` ssh target for remote · `-n` dst
+Key flags: `-L` named tmux socket · `-s` dst session (required) · `-H` ssh target for remote · `-n` dst
 hostname for the preamble (else auto-resolved) · `-m`/`-f`/stdin body · `-S`
 override source label · `-v` verbose · `-r N` Enter-flush attempts.
 
+For durable fleet use, prefer exact tmux targets such as `=coder0`. The helper
+normalizes exact session targets to pane-qualified targets internally so pane
+commands do not fall back to tmux's prefix matching behavior.
+
+## Named socket isolation
+
+Durable Mosaic fleets should use a dedicated tmux socket, for example:
+
+```bash
+tmux -L mosaic-factory ls
+agent-send.sh -L mosaic-factory -s '=coder0' -m "status?"
+send-message.sh -L mosaic-factory -t '=coder0' -m "raw pane message"
+```
+
+This keeps fleet operations away from the user's default tmux server. It is the
+safe rollout path on hosts that already have manual tmux sessions.
+
 ## Why a helper exists (the submission gotcha)
 
 Pasting into an interactive REPL via raw `tmux send-keys` is unreliable: a
@@ -67,6 +87,7 @@ message crosses the wire as base64 (`-b`) to avoid all shell-quoting hazards.
 
 - `agent-send.sh` — inter-agent wrapper (preamble + local/remote dispatch).
 - `send-message.sh` — low-level reliable single-pane submitter (`-b` base64 input).
+- `test-send-message-socket.sh` — smoke test for named-socket isolation.
 
 ## Distribution
 

packages/mosaic/framework/tools/tmux/agent-send.sh

@@ -12,10 +12,6 @@
 #   ambiguity about lanes or origin. Recipients replying should FLIP the
 #   preamble: [<dst> -> <src>] ... (this tool sends; it does not auto-reply).
 #
-#   Optionally tags the message with a TRIAGE CLASS (see -C / --class) so a
-#   comms daemon can route it (deliver-to-agent vs log-and-drop) from an exact
-#   field instead of re-deriving intent from the body.
-#
 # WHY A WRAPPER
 #   Reliable submission into an interactive REPL (Claude Code / Codex) is fiddly:
 #   a trailing Enter is often swallowed and the message sits as an unsubmitted
@@ -27,73 +23,40 @@
 #   the remote host; only bash + tmux + base64 (standard).
 #
 # USAGE
-#   agent-send.sh -s <dst_session> -m "message"                 # local target
-#   agent-send.sh -H user@host -s <dst_session> -m "message"    # remote target
-#   agent-send.sh -H user@host -n <dst_hostname> -s <sess> -f msg.txt
-#   agent-send.sh -s mos-claude --class terminal-log -m "ACK — received"
-#   echo "msg" | agent-send.sh -H user@host -s <dst_session>
+#   agent-send.sh [-L socket] -s <dst_session> -m "message"                 # local target
+#   agent-send.sh [-L socket] -H user@host -s <dst_session> -m "message"    # remote target
+#   agent-send.sh [-L socket] -H user@host -n <dst_hostname> -s <sess> -f msg.txt
+#   echo "msg" | agent-send.sh [-L socket] -H user@host -s <dst_session>
 #
 # OPTIONS
+#   -L NAME         tmux socket name passed to `tmux -L NAME` on the target host
 #   -s DST_SESSION  target tmux session (or session:window.pane)   [required]
 #   -H SSH_TARGET   ssh target (user@host) for a remote pane; omit for local
 #   -n DST_HOST     hostname to show in the preamble for the target.
 #                   Default: local hostname, or (remote) resolved via one ssh.
 #   -m MESSAGE      message text (single- or multi-line)
 #   -f FILE         read message from FILE instead of -m
-#   -C CLASS        triage class for a comms daemon. One of:
-#                     terminal-log  log-only; never needs the agent's attention
-#                     actionable    carries a decision/blocker/gate — deliver
-#                     human         from a human operator — deliver
-#                     reaction      an emoji/ack reaction
-#                   Long form: --class CLASS (or --class=CLASS). When SET, the
-#                   preamble carries a ` class=<CLASS>` token INSIDE the bracket:
-#                       [<src> -> <dst> class=terminal-log] <message>
-#                   When OMITTED, NO token is emitted and the preamble is
-#                   byte-for-byte identical to the classic format. Consumers MUST
-#                   treat an absent class as 'actionable' (fail-safe: agent sees it).
 #   -S SRC_LABEL    override source label "<host>:<session>" (default: auto)
 #   -r N            Enter-flush attempts passed through (default 2)
 #   -v              verbose: print pane tail after delivery
 #   -h              help
 #
-# PREAMBLE GRAMMAR  (for consumers / daemons mirroring this producer)
-#   ^\[(\S+) -> (\S+?)(?: class=(terminal-log|actionable|human|reaction))?\] (.*)$
-#     group 1 = src label   group 2 = dst host:session
-#     group 3 = class (absent => actionable)   group 4 = message body
-#
 # EXIT CODES  (passed through from send-message.sh)
 #   0 delivered/queued · 1 target not found · 2 still draft · 3 usage error
 set -uo pipefail
 
 SELF_DIR=$(cd -- "$(dirname -- "$0")" && pwd)
-# Sender is overridable via env purely for testing (inject a capture stub). The
-# default is the canonical send-message.sh beside this script; production callers
-# never set AGENT_SEND_SENDER, so behavior is unchanged.
-SENDER="${AGENT_SEND_SENDER:-$SELF_DIR/send-message.sh}"
+SENDER="$SELF_DIR/send-message.sh"
 
-# Translate the long option --class[=value] into "-C value" so getopts (which is
-# short-option-only) can parse it. Every other argument passes through untouched,
-# so callers that never use --class hit the exact original getopts path.
-args=()
-while [ $# -gt 0 ]; do
-  case "$1" in
-    --class)   [ $# -ge 2 ] || { echo "ERROR: --class requires a value" >&2; exit 3; }
-               args+=(-C "$2"); shift 2 ;;
-    --class=*) args+=(-C "${1#*=}"); shift ;;
-    *)         args+=("$1"); shift ;;
-  esac
-done
-set -- ${args[@]+"${args[@]}"}
+DST_SESSION=""; SSH_TARGET=""; DST_HOST=""; MSG=""; FILE=""; SOCKET_NAME=""
+SRC_LABEL=""; RETRIES=2; VERBOSE=0
+usage() { sed -n '2,44p' "$0"; exit "${1:-3}"; }
 
-DST_SESSION=""; SSH_TARGET=""; DST_HOST=""; MSG=""; FILE=""
-SRC_LABEL=""; RETRIES=2; VERBOSE=0; CLASS=""
-usage() { sed -n '2,/^set -uo pipefail/{/^set -uo pipefail/d;p}' "$0"; exit "${1:-3}"; }
-
-while getopts "s:H:n:m:f:S:r:C:vh" o; do
+while getopts "L:s:H:n:m:f:S:r:vh" o; do
   case "$o" in
+    L) SOCKET_NAME=$OPTARG ;;
     s) DST_SESSION=$OPTARG ;; H) SSH_TARGET=$OPTARG ;; n) DST_HOST=$OPTARG ;;
     m) MSG=$OPTARG ;; f) FILE=$OPTARG ;; S) SRC_LABEL=$OPTARG ;;
-    C) CLASS=$OPTARG ;;
     r) RETRIES=$OPTARG ;; v) VERBOSE=1 ;; h) usage 0 ;; *) usage 3 ;;
   esac
 done
@@ -101,17 +64,6 @@ done
 [ -n "$DST_SESSION" ] || { echo "ERROR: -s DST_SESSION is required" >&2; usage 3; }
 [ -x "$SENDER" ] || { echo "ERROR: send-message.sh not found beside this script" >&2; exit 3; }
 
-# Validate the triage class only when one was given. An absent class emits NO
-# token (preamble byte-identical to the classic format); the consumer defaults
-# absent => actionable.
-CLASS_TOKEN=""
-if [ -n "$CLASS" ]; then
-  case "$CLASS" in
-    terminal-log|actionable|human|reaction) CLASS_TOKEN=" class=${CLASS}" ;;
-    *) echo "ERROR: invalid --class '$CLASS' (allowed: terminal-log, actionable, human, reaction)" >&2; exit 3 ;;
-  esac
-fi
-
 # Message body from -f / -m / stdin.
 if   [ -n "$FILE" ]; then [ -r "$FILE" ] || { echo "ERROR: cannot read $FILE" >&2; exit 3; }; MSG=$(cat -- "$FILE")
 elif [ -z "$MSG" ] && [ ! -t 0 ]; then MSG=$(cat)
@@ -120,8 +72,12 @@ fi
 
 # Source label: this agent's host:session (auto-detected, overridable).
 if [ -z "$SRC_LABEL" ]; then
+  tmux_cmd=(tmux)
+  if [ -n "$SOCKET_NAME" ]; then
+    tmux_cmd+=(-L "$SOCKET_NAME")
+  fi
   src_host=$(hostname -s 2>/dev/null || echo "?")
-  src_sess=$(tmux display-message -p '#S' 2>/dev/null || echo "?")
+  src_sess=$("${tmux_cmd[@]}" display-message -p '#S' 2>/dev/null || echo "?")
   SRC_LABEL="${src_host}:${src_sess}"
 fi
 
@@ -134,17 +90,21 @@ if [ -z "$DST_HOST" ]; then
   fi
 fi
 
-PREAMBLE="[${SRC_LABEL} -> ${DST_HOST}:${DST_SESSION}${CLASS_TOKEN}]"
+PREAMBLE="[${SRC_LABEL} -> ${DST_HOST}:${DST_SESSION}]"
 FULL="${PREAMBLE} ${MSG}"
 B64=$(printf '%s' "$FULL" | base64 -w0)
 
 vflag=""; [ "$VERBOSE" = 1 ] && vflag="-v"
+socket_args=()
+if [ -n "$SOCKET_NAME" ]; then
+  socket_args=(-L "$SOCKET_NAME")
+fi
 
 if [ -z "$SSH_TARGET" ]; then
   # Local pane: call the canonical sender directly.
-  exec "$SENDER" -t "$DST_SESSION" -b "$B64" -r "$RETRIES" $vflag
+  exec "$SENDER" "${socket_args[@]}" -t "$DST_SESSION" -b "$B64" -r "$RETRIES" $vflag
 else
   # Remote pane: ship the sender over ssh and run it local to the target.
   ssh -o ConnectTimeout=10 "$SSH_TARGET" \
-      "bash -s -- -t '$DST_SESSION' -b '$B64' -r '$RETRIES' $vflag" < "$SENDER"
+      "bash -s -- ${socket_args[*]@Q} -t '$DST_SESSION' -b '$B64' -r '$RETRIES' $vflag" < "$SENDER"
 fi

packages/mosaic/framework/tools/tmux/agent-send.test.sh

@@ -1,97 +0,0 @@
-#!/usr/bin/env bash
-# agent-send.test.sh — regression + grammar lock for agent-send.sh --class.
-#
-# Strategy: inject a capture stub via AGENT_SEND_SENDER that decodes the -b
-# base64 payload and prints the FULL message (preamble + body) so we can assert
-# the exact bytes on the wire. Local path only (no ssh), -n pins the dst host so
-# the preamble is deterministic across machines.
-#
-# Guarantees locked here:
-#   1. REGRESSION BAR — no --class => preamble byte-for-byte identical to classic.
-#   2. --class <c> => ` class=<c>` token emitted inside the bracket.
-#   3. --class=<c> (equals form) parses identically to the space form.
-#   4. -C <c> short form parses identically.
-#   5. invalid class => exit 3, nothing sent.
-#   6. --class with no value => exit 3.
-#   7. the documented consumer regex parses producer output for every class.
-set -uo pipefail
-
-HERE=$(cd -- "$(dirname -- "$0")" && pwd)
-TOOL="$HERE/agent-send.sh"
-
-# Capture stub: stands in for send-message.sh. Decodes -b and prints the payload.
-STUB=$(mktemp)
-trap 'rm -f "$STUB"' EXIT
-cat >"$STUB" <<'STUB_EOF'
-#!/usr/bin/env bash
-set -uo pipefail
-b64=""
-while getopts "t:b:r:v" o; do case "$o" in b) b64=$OPTARG ;; *) : ;; esac; done
-printf '%s' "$b64" | base64 -d
-STUB_EOF
-chmod +x "$STUB"
-
-PASS=0; FAIL=0
-ok()   { PASS=$((PASS+1)); printf 'ok   %s\n' "$1"; }
-no()   { FAIL=$((FAIL+1)); printf 'FAIL %s\n   %s\n' "$1" "$2"; }
-
-# Run the tool with the stub injected; echoes captured payload on stdout.
-run() { AGENT_SEND_SENDER="$STUB" bash "$TOOL" -S a:src -n dsthost "$@"; }
-
-# Documented consumer grammar — the daemon will mirror exactly this.
-GRAMMAR='^\[(\S+) -> (\S+) class=(terminal-log|actionable|human|reaction)\] (.*)$'
-GRAMMAR_NOCLASS='^\[(\S+) -> (\S+)\] (.*)$'
-
-# 1. REGRESSION BAR: classic preamble, byte-for-byte.
-got=$(run -s mos -m "hello world")
-want='[a:src -> dsthost:mos] hello world'
-[ "$got" = "$want" ] && ok "regression: no --class is byte-identical" \
-  || no "regression: no --class is byte-identical" "got=[$got] want=[$want]"
-
-# 2. --class space form emits the token.
-got=$(run -s mos --class terminal-log -m "ACK")
-want='[a:src -> dsthost:mos class=terminal-log] ACK'
-[ "$got" = "$want" ] && ok "--class terminal-log emits token" \
-  || no "--class terminal-log emits token" "got=[$got] want=[$want]"
-
-# 3. --class=value equals form.
-got=$(run -s mos --class=actionable -m "decide X")
-want='[a:src -> dsthost:mos class=actionable] decide X'
-[ "$got" = "$want" ] && ok "--class=actionable (equals form)" \
-  || no "--class=actionable (equals form)" "got=[$got] want=[$want]"
-
-# 4. -C short form.
-got=$(run -s mos -C human -m "from a person")
-want='[a:src -> dsthost:mos class=human] from a person'
-[ "$got" = "$want" ] && ok "-C human (short form)" \
-  || no "-C human (short form)" "got=[$got] want=[$want]"
-
-# 5. invalid class => exit 3, no send.
-if out=$(run -s mos --class bogus -m "x" 2>/dev/null); then
-  no "invalid class rejected" "expected non-zero exit, got 0 (out=[$out])"
-else
-  rc=$?
-  [ "$rc" = 3 ] && [ -z "$out" ] && ok "invalid class => exit 3, nothing sent" \
-    || no "invalid class => exit 3, nothing sent" "rc=$rc out=[$out]"
-fi
-
-# 6. --class with no value => exit 3.
-if run -s mos -m "x" --class 2>/dev/null; then
-  no "--class with no value rejected" "expected non-zero exit, got 0"
-else
-  [ "$?" = 3 ] && ok "--class with no value => exit 3" || no "--class with no value => exit 3" "wrong rc"
-fi
-
-# 7. consumer grammar parses every class + classic line.
-for c in terminal-log actionable human reaction; do
-  line=$(run -s mos --class "$c" -m "body $c")
-  [[ "$line" =~ $GRAMMAR ]] && [ "${BASH_REMATCH[3]}" = "$c" ] && [ "${BASH_REMATCH[4]}" = "body $c" ] \
-    && ok "grammar parses class=$c" || no "grammar parses class=$c" "line=[$line]"
-done
-classic=$(run -s mos -m "plain body")
-[[ "$classic" =~ $GRAMMAR_NOCLASS ]] && [ "${BASH_REMATCH[3]}" = "plain body" ] \
-  && ok "grammar (no-class) parses classic line" || no "grammar (no-class) parses classic line" "line=[$classic]"
-
-echo "---"
-echo "PASS=$PASS FAIL=$FAIL"
-[ "$FAIL" -eq 0 ]

packages/mosaic/framework/tools/tmux/send-message.sh

@@ -13,12 +13,13 @@
 #   no-op in Claude Code, so the double-Enter is safe.
 #
 # USAGE
-#   send-message.sh -t <target> -m "message"
-#   send-message.sh -t <target> -f <file>
-#   echo "message" | send-message.sh -t <target>
-#   ssh host bash -s -- -t <target> -b "$(base64 -w0 <<<msg)" < send-message.sh
+#   send-message.sh [-L socket_name] -t <target> -m "message"
+#   send-message.sh [-L socket_name] -t <target> -f <file>
+#   echo "message" | send-message.sh [-L socket_name] -t <target>
+#   ssh host bash -s -- -L socket -t <target> -b "$(base64 -w0 <<<msg)" < send-message.sh
 #
 # OPTIONS
+#   -L NAME     tmux socket name passed to `tmux -L NAME` (optional)
 #   -t TARGET   tmux target: session, or session:window.pane  [required]
 #   -m MESSAGE  message text (single- or multi-line)
 #   -f FILE     read message from FILE instead of -m
@@ -34,11 +35,12 @@
 #   3  usage error
 set -uo pipefail
 
-TARGET=""; MSG=""; FILE=""; B64=""; RETRIES=2; VERBOSE=0
+SOCKET_NAME=""; TARGET=""; MSG=""; FILE=""; B64=""; RETRIES=2; VERBOSE=0
 usage() { sed -n '2,34p' "$0"; exit "${1:-3}"; }
 
-while getopts "t:m:f:b:r:vh" o; do
+while getopts "L:t:m:f:b:r:vh" o; do
   case "$o" in
+    L) SOCKET_NAME=$OPTARG ;;
     t) TARGET=$OPTARG ;; m) MSG=$OPTARG ;; f) FILE=$OPTARG ;; b) B64=$OPTARG ;;
     r) RETRIES=$OPTARG ;; v) VERBOSE=1 ;; h) usage 0 ;; *) usage 3 ;;
   esac
@@ -51,8 +53,21 @@ elif [ -z "$MSG" ] && [ ! -t 0 ]; then MSG=$(cat)
 fi
 [ -n "$MSG" ] || { echo "ERROR: empty message (use -m, -f, or stdin)" >&2; exit 3; }
 
+tmux_cmd=(tmux)
+if [ -n "$SOCKET_NAME" ]; then
+  tmux_cmd+=(-L "$SOCKET_NAME")
+fi
+
+# tmux accepts `=session` for some commands, but pane-level commands such as
+# capture-pane require a pane-qualified target. Keep exact-session addressing
+# convenient while avoiding accidental prefix matches.
+EFFECTIVE_TARGET=$TARGET
+if [[ "$TARGET" == =* && "$TARGET" != *:* ]]; then
+  EFFECTIVE_TARGET="${TARGET}:0.0"
+fi
+
 # Target must resolve to a live pane.
-if ! tmux list-panes -t "$TARGET" >/dev/null 2>&1; then
+if ! "${tmux_cmd[@]}" list-panes -t "$EFFECTIVE_TARGET" >/dev/null 2>&1; then
   echo "ERROR: tmux target not found: $TARGET" >&2; exit 1
 fi
 
@@ -62,18 +77,18 @@ snippet=$(printf '%s' "$MSG" | tr '\n' ' ' | tr -s ' ' | sed 's/[^[:print:]]//g'
 
 # 1) Paste the body as a bracketed paste so multi-line content does not submit
 #    line-by-line. load-buffer/paste-buffer is far safer than `send-keys -l`.
-printf '%s' "$MSG" | tmux load-buffer -b __mosaic_send -
+printf '%s' "$MSG" | "${tmux_cmd[@]}" load-buffer -b __mosaic_send -
 # -p = bracketed paste when the client supports it; fall back if not.
-tmux paste-buffer -d -p -b __mosaic_send -t "$TARGET" 2>/dev/null \
-  || tmux paste-buffer -d -b __mosaic_send -t "$TARGET"
+"${tmux_cmd[@]}" paste-buffer -d -p -b __mosaic_send -t "$EFFECTIVE_TARGET" 2>/dev/null \
+  || "${tmux_cmd[@]}" paste-buffer -d -b __mosaic_send -t "$EFFECTIVE_TARGET"
 sleep 0.5
 
 # 2) Submit, then verify; flush with another Enter if it is still a draft.
 status="sent"
 for attempt in $(seq 1 $((RETRIES + 1))); do
-  tmux send-keys -t "$TARGET" Enter
+  "${tmux_cmd[@]}" send-keys -t "$EFFECTIVE_TARGET" Enter
   sleep 1.2
-  pane=$(tmux capture-pane -t "$TARGET" -p 2>/dev/null)
+  pane=$("${tmux_cmd[@]}" capture-pane -t "$EFFECTIVE_TARGET" -p 2>/dev/null)
 
   if printf '%s' "$pane" | grep -qF "$QUEUED_RE"; then
     status="queued"; break

packages/mosaic/framework/tools/tmux/test-send-message-socket.sh

@@ -0,0 +1,50 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR=$(cd -- "$(dirname -- "$0")" && pwd)
+SEND_MESSAGE="$SCRIPT_DIR/send-message.sh"
+AGENT_SEND="$SCRIPT_DIR/agent-send.sh"
+SOCKET="mosaic-test-$RANDOM-$$"
+TARGET="target-$RANDOM"
+DEFAULT_TARGET="default-target-$RANDOM"
+TMPDIR=$(mktemp -d)
+trap 'tmux -L "$SOCKET" kill-server >/dev/null 2>&1 || true; tmux kill-session -t "$DEFAULT_TARGET" >/dev/null 2>&1 || true; rm -rf "$TMPDIR"' EXIT
+
+fail() {
+  echo "FAIL: $*" >&2
+  exit 1
+}
+
+require_tmux() {
+  command -v tmux >/dev/null 2>&1 || fail "tmux is required"
+}
+
+capture_named() {
+  tmux -L "$SOCKET" capture-pane -t "=$TARGET:0.0" -p
+}
+
+capture_default() {
+  tmux capture-pane -t "=$DEFAULT_TARGET:0.0" -p
+}
+
+require_tmux
+
+tmux -L "$SOCKET" new-session -d -s "$TARGET" -c "$TMPDIR" 'bash --noprofile --norc -i'
+tmux new-session -d -s "$DEFAULT_TARGET" -c "$TMPDIR" 'bash --noprofile --norc -i'
+
+"$SEND_MESSAGE" -L "$SOCKET" -t "=$TARGET" -m "named socket hello" >/tmp/send-message-named.out
+sleep 0.2
+capture_named | grep -qF "named socket hello" || fail "send-message.sh did not deliver to named socket"
+if capture_default | grep -qF "named socket hello"; then
+  fail "send-message.sh leaked named-socket message to default tmux server"
+fi
+
+"$AGENT_SEND" -L "$SOCKET" -S "tester:source" -s "=$TARGET" -m "agent socket hello" >/tmp/agent-send-named.out
+sleep 0.2
+capture_named | grep -qF "[tester:source ->" || fail "agent-send.sh did not include preamble"
+capture_named | grep -qF "agent socket hello" || fail "agent-send.sh did not deliver to named socket"
+if capture_default | grep -qF "agent socket hello"; then
+  fail "agent-send.sh leaked named-socket message to default tmux server"
+fi
+
+echo "ok - named tmux socket send tools"

packages/mosaic/framework/tools/woodpecker/README.md

@@ -26,11 +26,12 @@ A Woodpecker API token is required. To configure:
 
 ## Scripts
 
-| Script                | Purpose                                     |
-| --------------------- | ------------------------------------------- |
-| `pipeline-list.sh`    | List recent pipelines for a repo            |
-| `pipeline-status.sh`  | Get status of a specific or latest pipeline |
-| `pipeline-trigger.sh` | Trigger a new pipeline build                |
+| Script                | Purpose                                      |
+| --------------------- | -------------------------------------------- |
+| `pipeline-list.sh`    | List recent pipelines for a repo             |
+| `pipeline-status.sh`  | Get status of a specific or latest pipeline  |
+| `pipeline-trigger.sh` | Trigger a new pipeline build                 |
+| `ci-wait.sh`          | Block until pipeline(s) reach terminal state |
 
 ## Common Options
 
@@ -55,4 +56,7 @@ A Woodpecker API token is required. To configure:
 
 # Trigger a build on a specific branch
 ~/.config/mosaic/tools/woodpecker/pipeline-trigger.sh -b feature/my-branch
+
+# Block until one or more pipelines finish (event-driven CI wait)
+~/.config/mosaic/tools/woodpecker/ci-wait.sh -r usc/uconnect -n 3917 -n 3918
 ```

packages/mosaic/framework/tools/woodpecker/_lib.sh

@@ -12,7 +12,7 @@ wp_resolve_repo_id() {
   local full_name="$1"
   local response http_code body repo_id
 
-  response=$(curl -sk -w "\n%{http_code}" \
+  response=$(curl -sS -w "\n%{http_code}" \
     -H "Authorization: Bearer $WOODPECKER_TOKEN" \
     "${WOODPECKER_URL}/api/repos/lookup/${full_name}")
 

packages/mosaic/framework/tools/woodpecker/ci-wait.sh

@@ -0,0 +1,86 @@
+#!/usr/bin/env bash
+# ci-wait.sh — block until one or more Woodpecker pipelines reach terminal state.
+#
+# Problem it solves: orchestrators hand-author a `while true; curl .../repos/1/pipelines/$n
+# ...; sleep` loop for every CI wait. Those loops HARDCODE Woodpecker repo id 1 (only
+# correct for whichever repo happens to be id 1), re-implement URL building with raw
+# curl, and tend to get armed as tight <300s ScheduleWakeup polls (each poll = a full
+# wake+reload+recheck cycle). This encapsulates the loop once, on top of the existing
+# `pipeline-status.sh` wrapper (which resolves repo->id correctly and is instance-aware),
+# so a CI wait becomes a one-liner.
+#
+# Intended use: as the COMMAND of a Monitor / event-driven re-invoke (primary), paired
+# with a single long (>=1500s) timed fallback — NOT as a tight standalone poll.
+#
+# Usage:
+#   ci-wait.sh -r <owner/repo> -n <num> [-n <num> ...] [-a <instance>] [-i <interval>] [-t <timeout>]
+#   ci-wait.sh -r usc/uconnect -n 3917 -n 3918              # wait for both, infer instance
+#   ci-wait.sh -r usc/uconnect -n 3922 -a usc -i 30 -t 2400
+#
+# Instance is inferred from the owner (usc->usc, mosaicstack/mosaic->mosaic) unless -a given.
+# Exit: 0 = all pipelines terminal AND all 'success'; 1 = >=1 terminal non-success;
+#       2 = usage/precondition error; 3 = timeout before all terminal.
+set -euo pipefail
+
+# Resolve pipeline-status.sh as a sibling, matching how the woodpecker tools source
+# _lib.sh — works under the installed runtime AND an in-repo checkout, no MOSAIC_HOME dep.
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PS="$SCRIPT_DIR/pipeline-status.sh"
+
+REPO="" INSTANCE="" INTERVAL=30 TIMEOUT=3600
+NUMS=()
+while getopts "r:n:a:i:t:h" opt; do
+  case "$opt" in
+    r) REPO="$OPTARG" ;;
+    n) NUMS+=("$OPTARG") ;;
+    a) INSTANCE="$OPTARG" ;;
+    i) INTERVAL="$OPTARG" ;;
+    t) TIMEOUT="$OPTARG" ;;
+    h) grep '^#' "$0" | sed 's/^# \?//'; exit 0 ;;
+    *) echo "see -h" >&2; exit 2 ;;
+  esac
+done
+[[ -n "$REPO" ]] || { echo "FATAL: -r <owner/repo> required" >&2; exit 2; }
+[[ ${#NUMS[@]} -gt 0 ]] || { echo "FATAL: at least one -n <pipeline-number> required" >&2; exit 2; }
+[[ -x "$PS" ]] || { echo "FATAL: pipeline-status.sh not found/executable at $PS" >&2; exit 2; }
+
+# Infer Woodpecker instance from owner unless overridden (matches the git-wrapper convention).
+if [[ -z "$INSTANCE" ]]; then
+  case "${REPO%%/*}" in
+    usc|USC) INSTANCE=usc ;;
+    mosaicstack|mosaic) INSTANCE=mosaic ;;
+    *) echo "FATAL: cannot infer Woodpecker instance for owner '${REPO%%/*}' — pass -a <instance>" >&2; exit 2 ;;
+  esac
+fi
+
+command -v jq >/dev/null || { echo "FATAL: jq not found" >&2; exit 2; }
+
+TERMINAL_RE='^(success|failure|error|killed|declined|blocked)$'
+declare -A STATE=()        # num -> terminal status, once reached
+start=$(date +%s 2>/dev/null || echo 0)
+
+echo "ci-wait: $REPO pipelines [${NUMS[*]}] (instance=$INSTANCE, every ${INTERVAL}s, timeout ${TIMEOUT}s)"
+while true; do
+  for n in "${NUMS[@]}"; do
+    [[ -n "${STATE[$n]:-}" ]] && continue
+    s=$("$PS" -r "$REPO" -n "$n" -a "$INSTANCE" -f json 2>/dev/null | jq -r '.status // empty' 2>/dev/null || true)
+    if [[ "$s" =~ $TERMINAL_RE ]]; then
+      STATE[$n]="$s"
+      echo "  pipeline $n TERMINAL: $s"
+    fi
+  done
+  # all terminal?
+  if [[ ${#STATE[@]} -eq ${#NUMS[@]} ]]; then
+    bad=0
+    for n in "${NUMS[@]}"; do [[ "${STATE[$n]}" == "success" ]] || bad=1; done
+    if [[ $bad -eq 0 ]]; then echo "ci-wait: ALL SUCCESS"; exit 0; fi
+    echo "ci-wait: all terminal, NOT all success — $(for n in "${NUMS[@]}"; do printf '%s=%s ' "$n" "${STATE[$n]}"; done)"
+    exit 1
+  fi
+  now=$(date +%s 2>/dev/null || echo 0)
+  if [[ "$start" != 0 && $((now - start)) -ge $TIMEOUT ]]; then
+    echo "ci-wait: TIMEOUT after ${TIMEOUT}s — pending: $(for n in "${NUMS[@]}"; do [[ -z "${STATE[$n]:-}" ]] && printf '%s ' "$n"; done)"
+    exit 3
+  fi
+  sleep "$INTERVAL"
+done

packages/mosaic/framework/tools/woodpecker/pipeline-list.sh

@@ -48,7 +48,7 @@ fi
 # Resolve owner/repo to numeric ID (Woodpecker v3 API)
 REPO_ID=$(wp_resolve_repo_id "$REPO") || exit 1
 
-response=$(curl -sk -w "\n%{http_code}" \
+response=$(curl -sS -w "\n%{http_code}" \
   -H "Authorization: Bearer $WOODPECKER_TOKEN" \
   "${WOODPECKER_URL}/api/repos/${REPO_ID}/pipelines?perPage=${LIMIT}")
 

packages/mosaic/framework/tools/woodpecker/pipeline-status.sh

@@ -50,7 +50,7 @@ REPO_ID=$(wp_resolve_repo_id "$REPO") || exit 1
 _wp_fetch() {
   local ep="$1"
   local resp http_code body
-  resp=$(curl -sk -w "\n%{http_code}" \
+  resp=$(curl -sS -w "\n%{http_code}" \
     -H "Authorization: Bearer $WOODPECKER_TOKEN" \
     "$ep")
   http_code=$(echo "$resp" | tail -n1)

packages/mosaic/framework/tools/woodpecker/pipeline-trigger.sh

@@ -46,7 +46,7 @@ REPO_ID=$(wp_resolve_repo_id "$REPO") || exit 1
 
 echo "Triggering pipeline for $REPO on branch $BRANCH..."
 
-response=$(curl -sk -w "\n%{http_code}" -X POST \
+response=$(curl -sS -w "\n%{http_code}" -X POST \
   -H "Authorization: Bearer $WOODPECKER_TOKEN" \
   -H "Content-Type: application/json" \
   -d "$(jq -n --arg b "$BRANCH" '{branch: $b}')" \

packages/mosaic/framework/tools/woodpecker/test-ci-wait-exit-matrix.sh

@@ -0,0 +1,76 @@
+#!/usr/bin/env bash
+# Regression harness for ci-wait.sh terminal-state aggregation and exit codes.
+#
+# ci-wait.sh wraps pipeline-status.sh and blocks until every requested pipeline
+# reaches a terminal Woodpecker state, then maps the aggregate to an exit code.
+# That contract is what callers arm a Monitor/timed-fallback around, so it must be
+# exact. This harness drives ci-wait.sh against a stub pipeline-status.sh whose
+# per-pipeline status is fixture-controlled, and asserts the full exit matrix:
+#
+#   0 = every pipeline terminal AND all 'success'
+#   1 = every pipeline terminal, at least one non-success
+#   2 = usage/precondition error (missing -n)
+#   3 = timeout before all pipelines terminal
+#
+# Non-vacuity: each case pins a DISTINCT exit code to a distinct fixture, so a
+# regression in success-aggregation (case 0 vs 1), terminal detection (case 3),
+# or arg validation (case 2) flips exactly one assertion RED.
+
+set -euo pipefail
+
+CIW_SRC="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)/ci-wait.sh"
+WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/ci-wait-exit-matrix}"
+TOOL_DIR="$WORK_DIR/tool"
+
+rm -rf "$WORK_DIR"
+mkdir -p "$TOOL_DIR"
+
+# ci-wait.sh resolves pipeline-status.sh as a sibling ($SCRIPT_DIR/pipeline-status.sh),
+# so we run a COPY of ci-wait.sh next to a stub sibling we control.
+cp "$CIW_SRC" "$TOOL_DIR/ci-wait.sh"
+chmod +x "$TOOL_DIR/ci-wait.sh"
+
+# Stub pipeline-status.sh: emits {"status":"<s>"} where <s> comes from env
+# CIW_STATUS_<num> (default "running" = non-terminal, drives the timeout path).
+cat > "$TOOL_DIR/pipeline-status.sh" <<'SH'
+#!/usr/bin/env bash
+set -euo pipefail
+num=""
+while getopts "r:n:a:f:" opt; do case "$opt" in n) num="$OPTARG" ;; *) : ;; esac; done
+var="CIW_STATUS_${num}"
+printf '{"status":"%s"}\n' "${!var:-running}"
+SH
+chmod +x "$TOOL_DIR/pipeline-status.sh"
+
+CIW="$TOOL_DIR/ci-wait.sh"
+
+run_expect() { # $1 = expected exit  $2 = label ; rest = args
+  local want="$1" label="$2"; shift 2
+  local rc=0
+  "$CIW" "$@" >/dev/null 2>&1 || rc=$?
+  if [[ "$rc" -ne "$want" ]]; then
+    echo "FAIL [$label]: expected exit $want, got $rc" >&2; exit 1
+  fi
+  echo "PASS [$label]: exit $rc"
+}
+
+# 0 — both pipelines terminal + success
+CIW_STATUS_100=success CIW_STATUS_101=success \
+  run_expect 0 "all-success" -r mosaic/stack -n 100 -n 101 -a mosaic -i 1 -t 30
+
+# 1 — both terminal, one failure
+CIW_STATUS_100=success CIW_STATUS_101=failure \
+  run_expect 1 "terminal-not-success" -r mosaic/stack -n 100 -n 101 -a mosaic -i 1 -t 30
+
+# 1 — other terminal non-success states still map to 1 (error/killed)
+CIW_STATUS_100=error CIW_STATUS_101=killed \
+  run_expect 1 "terminal-error-killed" -r mosaic/stack -n 100 -n 101 -a mosaic -i 1 -t 30
+
+# 3 — a pipeline never reaches terminal state before timeout
+CIW_STATUS_100=success CIW_STATUS_101=running \
+  run_expect 3 "timeout-pending" -r mosaic/stack -n 100 -n 101 -a mosaic -i 1 -t 0
+
+# 2 — usage error: no -n
+run_expect 2 "usage-missing-n" -r mosaic/stack -a mosaic
+
+echo "ALL PASS: test-ci-wait-exit-matrix.sh"

packages/mosaic/src/cli.ts

@@ -13,6 +13,7 @@ import { registerStorageCommand } from '@mosaicstack/storage';
 import { registerTelemetryCommand } from './commands/telemetry.js';
 import { registerAgentCommand } from './commands/agent.js';
 import { registerConfigCommand } from './commands/config.js';
+import { registerFleetCommand } from './commands/fleet.js';
 import { registerMissionCommand } from './commands/mission.js';
 import { registerUninstallCommand } from './commands/uninstall.js';
 // prdy is registered via launch.ts
@@ -57,7 +58,7 @@ Command Groups:
 
   Runtime:    tui, login, sessions
   Gateway:    gateway
-  Framework:  agent, bootstrap, coord, doctor, init, launch, mission, prdy, seq, sync, upgrade, wizard, yolo
+  Framework:  agent, bootstrap, coord, doctor, fleet, init, launch, mission, prdy, seq, sync, upgrade, wizard, yolo
   Platform:   update
   Runtimes:   claude, codex, opencode, pi
 `,
@@ -345,6 +346,10 @@ registerFederationCommand(program);
 
 registerAgentCommand(program);
 
+// ─── fleet ─────────────────────────────────────────────────────────────
+
+registerFleetCommand(program);
+
 // ─── config ────────────────────────────────────────────────────────────
 
 registerConfigCommand(program);

packages/mosaic/src/commands/agent.ts

@@ -1,4 +1,5 @@
 import type { Command } from 'commander';
+import { registerFleetAgentCommands, type FleetCommandDeps } from './fleet.js';
 import { withAuth } from './with-auth.js';
 import { selectItem } from './select-dialog.js';
 import {
@@ -30,11 +31,13 @@ function showAgentDetail(a: AgentConfigInfo) {
   console.log(`  Created:      ${new Date(a.createdAt).toLocaleString()}`);
 }
 
-export function registerAgentCommand(program: Command) {
+export function registerAgentCommand(program: Command, fleetDeps: FleetCommandDeps = {}) {
   const cmd = program
     .command('agent')
-    .description('Manage agent configurations')
+    .description('Manage agent configurations and local fleet agents')
     .option('-g, --gateway <url>', 'Gateway URL', 'http://localhost:14242')
+    .option('--mosaic-home <path>', 'Mosaic home directory')
+    .option('--roster <path>', 'Local fleet roster path')
     .option('--list', 'List all agents')
     .option('--new', 'Create a new agent')
     .option('--show <idOrName>', 'Show agent details')
@@ -72,6 +75,8 @@ export function registerAgentCommand(program: Command) {
       },
     );
 
+  registerFleetAgentCommands(cmd, fleetDeps);
+
   return cmd;
 }
 

packages/mosaic/src/commands/fleet.spec.ts

@@ -0,0 +1,563 @@
+import { mkdir, mkdtemp, readFile, rm, writeFile } from 'node:fs/promises';
+import { tmpdir } from 'node:os';
+import { dirname, join, resolve } from 'node:path';
+import { Command } from 'commander';
+import { afterEach, describe, expect, it, vi } from 'vitest';
+import {
+  buildAgentSendCommand,
+  buildFleetServiceCommand,
+  generateAgentEnv,
+  getRosterAgent,
+  loadFleetRoster,
+  registerFleetCommand,
+  resolveFleetPaths,
+  type CommandRunner,
+} from './fleet.js';
+import { registerAgentCommand } from './agent.js';
+
+function buildProgram(): Command {
+  const program = new Command();
+  program.exitOverride();
+  registerFleetCommand(program);
+  registerAgentCommand(program);
+  return program;
+}
+
+async function tempDir(): Promise<string> {
+  return mkdtemp(join(tmpdir(), 'mosaic-fleet-'));
+}
+
+describe('registerFleetCommand', () => {
+  it('registers local canary fleet subcommands', () => {
+    const program = buildProgram();
+    const fleet = program.commands.find((command) => command.name() === 'fleet');
+
+    expect(fleet).toBeDefined();
+    expect(fleet!.commands.map((command) => command.name()).sort()).toEqual([
+      'init',
+      'install',
+      'install-systemd',
+      'restart',
+      'start',
+      'status',
+      'stop',
+      'verify',
+    ]);
+  });
+
+  it('adds fleet-backed agent subcommands without removing existing options', () => {
+    const program = buildProgram();
+    const agent = program.commands.find((command) => command.name() === 'agent');
+
+    expect(agent).toBeDefined();
+    expect(agent!.options.map((option) => option.long)).toContain('--list');
+    expect(agent!.commands.map((command) => command.name()).sort()).toEqual([
+      'reset',
+      'roster',
+      'send',
+      'status',
+      'tail',
+    ]);
+  });
+});
+
+describe('fleet roster parsing', () => {
+  let cleanup: string | undefined;
+
+  afterEach(async () => {
+    if (cleanup) {
+      await rm(cleanup, { recursive: true, force: true });
+      cleanup = undefined;
+    }
+  });
+
+  it('defaults local canary rosters to the isolated mosaic-factory socket', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.yaml');
+    await writeFile(
+      rosterPath,
+      [
+        'version: 1',
+        'transport: tmux',
+        'agents:',
+        '  - name: canary-pi',
+        '    runtime: pi',
+        '    class: canary',
+      ].join('\n'),
+    );
+
+    const roster = await loadFleetRoster(rosterPath);
+
+    expect(roster.tmux.socketName).toBe('mosaic-factory');
+    expect(roster.tmux.holderSession).toBe('_holder');
+    expect(roster.agents).toHaveLength(1);
+    expect(getRosterAgent(roster, 'canary-pi').runtime).toBe('pi');
+  });
+
+  it('generates deterministic per-agent EnvironmentFile content', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.json');
+    await writeFile(
+      rosterPath,
+      JSON.stringify({
+        version: 1,
+        transport: 'tmux',
+        tmux: { socket_name: 'mosaic-factory' },
+        defaults: { working_directory: '/srv/mosaic' },
+        agents: [{ name: 'coder0', runtime: 'codex', class: 'implementer' }],
+      }),
+    );
+    const roster = await loadFleetRoster(rosterPath);
+
+    expect(generateAgentEnv(roster, getRosterAgent(roster, 'coder0'))).toBe(
+      [
+        'MOSAIC_AGENT_NAME=coder0',
+        'MOSAIC_AGENT_RUNTIME=codex',
+        'MOSAIC_AGENT_WORKDIR=/srv/mosaic',
+        'MOSAIC_TMUX_SOCKET=mosaic-factory',
+        '',
+      ].join('\n'),
+    );
+  });
+
+  it('rejects unknown roster fields instead of silently defaulting', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.yaml');
+    await writeFile(
+      rosterPath,
+      [
+        'version: 1',
+        'transport: tmux',
+        'tmux:',
+        '  socketNamee: prod-fleet',
+        'agents:',
+        '  - name: canary-pi',
+        '    runtime: pi',
+      ].join('\n'),
+    );
+
+    await expect(loadFleetRoster(rosterPath)).rejects.toThrow(
+      'Fleet roster tmux has unknown field(s): socketNamee.',
+    );
+  });
+
+  it('rejects wrong-typed roster fields instead of silently defaulting', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.json');
+    await writeFile(
+      rosterPath,
+      JSON.stringify({
+        version: 1,
+        transport: 'tmux',
+        tmux: { socket_name: 123 },
+        defaults: { working_directory: '/srv/mosaic' },
+        agents: [{ name: 'canary-pi', runtime: 'pi' }],
+      }),
+    );
+
+    await expect(loadFleetRoster(rosterPath)).rejects.toThrow(
+      'Fleet roster tmux socket_name must be a string.',
+    );
+  });
+
+  it('rejects wrong-typed agent fields', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.json');
+    await writeFile(
+      rosterPath,
+      JSON.stringify({
+        version: 1,
+        transport: 'tmux',
+        agents: [{ name: 'canary-pi', runtime: 42 }],
+      }),
+    );
+
+    await expect(loadFleetRoster(rosterPath)).rejects.toThrow(
+      'Fleet roster agent "canary-pi" runtime must be a string.',
+    );
+  });
+
+  it('rejects duplicate agent names before install can overwrite env files', async () => {
+    cleanup = await tempDir();
+    const rosterPath = join(cleanup, 'roster.yaml');
+    await writeFile(
+      rosterPath,
+      [
+        'version: 1',
+        'transport: tmux',
+        'agents:',
+        '  - name: canary-pi',
+        '    runtime: pi',
+        '  - name: canary-pi',
+        '    runtime: codex',
+      ].join('\n'),
+    );
+
+    await expect(loadFleetRoster(rosterPath)).rejects.toThrow(
+      'Fleet roster has duplicate agent name: canary-pi.',
+    );
+  });
+
+  it('ships generic minimal and local-canary examples without site-specific defaults', async () => {
+    const examplesDir = resolve(process.cwd(), 'framework', 'fleet', 'examples');
+    const minimal = await loadFleetRoster(join(examplesDir, 'minimal.yaml'));
+    const localCanaryText = await readFile(join(examplesDir, 'local-canary.yaml'), 'utf8');
+    const localCanary = await loadFleetRoster(join(examplesDir, 'local-canary.yaml'));
+
+    expect(minimal.agents.map((agent) => agent.name)).toEqual(['canary-pi']);
+    expect(localCanary.tmux.socketName).toBe('mosaic-factory');
+    expect(localCanary.agents.map((agent) => agent.name)).toEqual(['lead', 'coder0', 'reviewer0']);
+    expect(localCanaryText).not.toMatch(/usc|ultron|secrev/i);
+  });
+});
+
+describe('fleet command construction', () => {
+  it('builds exact systemd user commands for holder and agent operations', () => {
+    expect(buildFleetServiceCommand('status')).toEqual([
+      'systemctl',
+      '--user',
+      'status',
+      'mosaic-tmux-holder.service',
+    ]);
+    expect(buildFleetServiceCommand('restart', 'coder0')).toEqual([
+      'systemctl',
+      '--user',
+      'restart',
+      'mosaic-agent@coder0.service',
+    ]);
+  });
+
+  it('builds socket-scoped agent send commands', () => {
+    const paths = resolveFleetPaths('/home/test/.config/mosaic');
+    expect(buildAgentSendCommand(paths, 'coder0', 'hello', 'mosaic-factory')).toEqual([
+      '/home/test/.config/mosaic/tools/tmux/agent-send.sh',
+      '-L',
+      'mosaic-factory',
+      '-s',
+      'coder0',
+      '-m',
+      'hello',
+    ]);
+  });
+
+  it('runs fleet status through injected runner without touching tmux in tests', async () => {
+    const calls: string[][] = [];
+    const runner: CommandRunner = async (command, args) => {
+      calls.push([command, ...args]);
+      return { stdout: 'ok\n', stderr: '', exitCode: 0 };
+    };
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner });
+
+    await program.parseAsync(['node', 'mosaic', 'fleet', 'status']);
+
+    expect(calls).toEqual([['systemctl', '--user', 'status', 'mosaic-tmux-holder.service']]);
+  });
+
+  it('writes init output to the explicit roster path', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'custom', 'roster.yaml');
+    const frameworkRoot = resolve(process.cwd(), 'framework');
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { frameworkRoot, mosaicHome: home });
+
+    try {
+      await program.parseAsync([
+        'node',
+        'mosaic',
+        'fleet',
+        '--roster',
+        rosterPath,
+        'init',
+        '--profile',
+        'minimal',
+        '--write',
+      ]);
+
+      const content = await readFile(rosterPath, 'utf8');
+      expect(content).toContain('name: canary-pi');
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('refuses to overwrite an existing roster unless --force is provided', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'custom', 'roster.yaml');
+    await mkdir(dirname(rosterPath), { recursive: true });
+    await writeFile(rosterPath, 'site-owned: true\n');
+    const frameworkRoot = resolve(process.cwd(), 'framework');
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { frameworkRoot, mosaicHome: home });
+
+    try {
+      await expect(
+        program.parseAsync([
+          'node',
+          'mosaic',
+          'fleet',
+          '--roster',
+          rosterPath,
+          'init',
+          '--profile',
+          'minimal',
+          '--write',
+        ]),
+      ).rejects.toThrow('Fleet roster already exists');
+      expect(await readFile(rosterPath, 'utf8')).toBe('site-owned: true\n');
+
+      await program.parseAsync([
+        'node',
+        'mosaic',
+        'fleet',
+        '--roster',
+        rosterPath,
+        'init',
+        '--profile',
+        'minimal',
+        '--write',
+        '--force',
+      ]);
+      expect(await readFile(rosterPath, 'utf8')).toContain('name: canary-pi');
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('rejects unknown init profiles instead of silently falling back', async () => {
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { frameworkRoot: resolve(process.cwd(), 'framework') });
+
+    await expect(
+      program.parseAsync(['node', 'mosaic', 'fleet', 'init', '--profile', 'typo']),
+    ).rejects.toThrow('Unsupported fleet profile');
+  });
+
+  it('sets process exitCode when status runner fails', async () => {
+    const originalExitCode = process.exitCode;
+    const stderrSpy = vi.spyOn(process.stderr, 'write').mockImplementation(() => true);
+    const runner: CommandRunner = async () => ({ stdout: '', stderr: 'missing\n', exitCode: 3 });
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner });
+
+    try {
+      await program.parseAsync(['node', 'mosaic', 'fleet', 'status']);
+      expect(process.exitCode).toBe(3);
+    } finally {
+      process.exitCode = originalExitCode;
+      stderrSpy.mockRestore();
+    }
+  });
+
+  it('loads default fleet/roster.json when roster.yaml is absent', async () => {
+    const home = await tempDir();
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      join(home, 'fleet', 'roster.json'),
+      JSON.stringify({
+        version: 1,
+        transport: 'tmux',
+        agents: [{ name: 'json-canary', runtime: 'pi' }],
+      }),
+    );
+    const calls: string[][] = [];
+    const runner: CommandRunner = async (command, args) => {
+      calls.push([command, ...args]);
+      return { stdout: '', stderr: '', exitCode: 0 };
+    };
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner, mosaicHome: home });
+
+    try {
+      await program.parseAsync(['node', 'mosaic', 'fleet', 'status', 'json-canary']);
+      expect(calls).toEqual([
+        ['systemctl', '--user', 'status', 'mosaic-agent@json-canary.service'],
+      ]);
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('starts the holder before agents and stops agents before the holder', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'fleet', 'roster.yaml');
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      rosterPath,
+      ['version: 1', 'transport: tmux', 'agents:', '  - name: coder0', '    runtime: codex'].join(
+        '\n',
+      ),
+    );
+    const calls: string[][] = [];
+    const runner: CommandRunner = async (command, args) => {
+      calls.push([command, ...args]);
+      return { stdout: '', stderr: '', exitCode: 0 };
+    };
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner, mosaicHome: home });
+
+    try {
+      await program.parseAsync(['node', 'mosaic', 'fleet', 'start']);
+      await program.parseAsync(['node', 'mosaic', 'fleet', 'stop']);
+
+      expect(calls).toEqual([
+        ['systemctl', '--user', 'start', 'mosaic-tmux-holder.service'],
+        ['systemctl', '--user', 'start', 'mosaic-agent@coder0.service'],
+        ['systemctl', '--user', 'stop', 'mosaic-agent@coder0.service'],
+        ['systemctl', '--user', 'stop', 'mosaic-tmux-holder.service'],
+      ]);
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('attempts every agent and the holder during fleet stop even when an agent stop fails', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'fleet', 'roster.yaml');
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      rosterPath,
+      [
+        'version: 1',
+        'transport: tmux',
+        'agents:',
+        '  - name: coder0',
+        '    runtime: codex',
+        '  - name: reviewer0',
+        '    runtime: pi',
+      ].join('\n'),
+    );
+    const calls: string[][] = [];
+    const runner: CommandRunner = async (command, args) => {
+      calls.push([command, ...args]);
+      if (args.includes('mosaic-agent@coder0.service')) {
+        return { stdout: '', stderr: 'coder0 failed\n', exitCode: 1 };
+      }
+      return { stdout: '', stderr: '', exitCode: 0 };
+    };
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner, mosaicHome: home });
+
+    try {
+      await expect(program.parseAsync(['node', 'mosaic', 'fleet', 'stop'])).rejects.toThrow(
+        'Fleet stop completed with 1 failure(s)',
+      );
+      expect(calls).toEqual([
+        ['systemctl', '--user', 'stop', 'mosaic-agent@coder0.service'],
+        ['systemctl', '--user', 'stop', 'mosaic-agent@reviewer0.service'],
+        ['systemctl', '--user', 'stop', 'mosaic-tmux-holder.service'],
+      ]);
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('rejects install-systemd with a non-default Mosaic home because units use %h/.config/mosaic', async () => {
+    const home = await tempDir();
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, {
+      mosaicHome: home,
+      frameworkRoot: resolve(process.cwd(), 'framework'),
+    });
+
+    try {
+      await expect(
+        program.parseAsync(['node', 'mosaic', 'fleet', 'install-systemd']),
+      ).rejects.toThrow('install-systemd only supports the default Mosaic home');
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it.each(['start', 'stop', 'restart', 'status'] as const)(
+    'rejects single-agent %s for agents outside the roster',
+    async (action) => {
+      const home = await tempDir();
+      const rosterPath = join(home, 'fleet', 'roster.yaml');
+      await mkdir(join(home, 'fleet'), { recursive: true });
+      await writeFile(
+        rosterPath,
+        ['version: 1', 'transport: tmux', 'agents:', '  - name: coder0', '    runtime: codex'].join(
+          '\n',
+        ),
+      );
+      const runner = vi.fn<CommandRunner>(async () => ({ stdout: '', stderr: '', exitCode: 0 }));
+      const program = new Command();
+      program.exitOverride();
+      registerFleetCommand(program, { runner, mosaicHome: home });
+
+      try {
+        await expect(
+          program.parseAsync(['node', 'mosaic', 'fleet', action, 'typo']),
+        ).rejects.toThrow('Agent "typo" is not in the fleet roster');
+        expect(runner).not.toHaveBeenCalled();
+      } finally {
+        await rm(home, { recursive: true, force: true });
+      }
+    },
+  );
+
+  it('loads default fleet/roster.json for agent commands when roster.yaml is absent', async () => {
+    const home = await tempDir();
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      join(home, 'fleet', 'roster.json'),
+      JSON.stringify({
+        version: 1,
+        transport: 'tmux',
+        agents: [{ name: 'json-agent', runtime: 'pi' }],
+      }),
+    );
+    const calls: string[][] = [];
+    const runner: CommandRunner = async (command, args) => {
+      calls.push([command, ...args]);
+      return { stdout: '', stderr: '', exitCode: 0 };
+    };
+    const program = new Command();
+    program.exitOverride();
+    registerAgentCommand(program, { runner, mosaicHome: home });
+
+    try {
+      await program.parseAsync(['node', 'mosaic', 'agent', 'status', 'json-agent']);
+      expect(calls).toEqual([
+        ['tmux', '-L', 'mosaic-factory', 'has-session', '-t', '=json-agent:0.0'],
+      ]);
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('rejects agent status typos before invoking the runner', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'fleet', 'roster.yaml');
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      rosterPath,
+      ['version: 1', 'transport: tmux', 'agents:', '  - name: coder0', '    runtime: codex'].join(
+        '\n',
+      ),
+    );
+    const runner = vi.fn<CommandRunner>(async () => ({ stdout: '', stderr: '', exitCode: 0 }));
+    const program = new Command();
+    program.exitOverride();
+    registerAgentCommand(program, { runner, mosaicHome: home });
+
+    try {
+      await expect(
+        program.parseAsync(['node', 'mosaic', 'agent', 'status', 'typo']),
+      ).rejects.toThrow('Agent "typo" is not in the fleet roster');
+      expect(runner).not.toHaveBeenCalled();
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+});

packages/mosaic/src/commands/fleet.ts

@@ -0,0 +1,851 @@
+import { constants } from 'node:fs';
+import { access, copyFile, mkdir, readFile, writeFile } from 'node:fs/promises';
+import { homedir } from 'node:os';
+import { dirname, join, resolve } from 'node:path';
+import { fileURLToPath } from 'node:url';
+import { spawn } from 'node:child_process';
+import type { Command } from 'commander';
+import YAML from 'yaml';
+
+export interface CommandResult {
+  stdout: string;
+  stderr: string;
+  exitCode: number;
+}
+
+export type CommandRunner = (command: string, args: string[]) => Promise<CommandResult>;
+
+export interface FleetCommandDeps {
+  runner?: CommandRunner;
+  mosaicHome?: string;
+  frameworkRoot?: string;
+}
+
+interface RawFleetRoster {
+  version?: unknown;
+  transport?: unknown;
+  tmux?: {
+    socket_name?: unknown;
+    socketName?: unknown;
+    holder_session?: unknown;
+    holderSession?: unknown;
+  };
+  defaults?: {
+    working_directory?: unknown;
+    workingDirectory?: unknown;
+  };
+  runtimes?: Record<string, { reset_command?: unknown; resetCommand?: unknown }>;
+  agents?: Array<{
+    name?: unknown;
+    runtime?: unknown;
+    class?: unknown;
+    working_directory?: unknown;
+    workingDirectory?: unknown;
+    model_hint?: unknown;
+    modelHint?: unknown;
+    persistent_persona?: unknown;
+    persistentPersona?: unknown;
+    reset_between_tasks?: unknown;
+    resetBetweenTasks?: unknown;
+    kickstart_template?: unknown;
+    kickstartTemplate?: unknown;
+  }>;
+}
+
+export interface FleetAgent {
+  name: string;
+  runtime: string;
+  className: string;
+  workingDirectory?: string;
+  modelHint?: string;
+  persistentPersona?: boolean | string;
+  resetBetweenTasks?: boolean;
+  kickstartTemplate?: string;
+}
+
+export interface FleetRoster {
+  version: 1;
+  transport: 'tmux';
+  tmux: {
+    socketName: string;
+    holderSession: string;
+  };
+  defaults: {
+    workingDirectory: string;
+  };
+  runtimes: Record<string, { resetCommand: string }>;
+  agents: FleetAgent[];
+}
+
+export interface FleetPaths {
+  mosaicHome: string;
+  rosterPath: string;
+  toolsDir: string;
+  fleetToolsDir: string;
+  tmuxToolsDir: string;
+  systemdUserDir: string;
+  agentEnvDir: string;
+}
+
+type FleetServiceAction = 'start' | 'stop' | 'restart' | 'status';
+
+const DEFAULT_SOCKET_NAME = 'mosaic-factory';
+const DEFAULT_HOLDER_SESSION = '_holder';
+const DEFAULT_WORKING_DIRECTORY = '~/src';
+const DEFAULT_RUNTIME_RESETS: Record<string, { resetCommand: string }> = {
+  claude: { resetCommand: '/clear' },
+  codex: { resetCommand: '/clear' },
+  opencode: { resetCommand: '/clear' },
+  pi: { resetCommand: '/new' },
+};
+
+export function resolveFleetPaths(mosaicHome = defaultMosaicHome()): FleetPaths {
+  return {
+    mosaicHome,
+    rosterPath: join(mosaicHome, 'fleet', 'roster.yaml'),
+    toolsDir: join(mosaicHome, 'tools'),
+    fleetToolsDir: join(mosaicHome, 'tools', 'fleet'),
+    tmuxToolsDir: join(mosaicHome, 'tools', 'tmux'),
+    systemdUserDir: join(homedir(), '.config', 'systemd', 'user'),
+    agentEnvDir: join(mosaicHome, 'fleet', 'agents'),
+  };
+}
+
+function defaultMosaicHome(): string {
+  return join(homedir(), '.config', 'mosaic');
+}
+
+function assertDefaultMosaicHomeForSystemd(mosaicHome: string): void {
+  if (resolve(mosaicHome) !== resolve(defaultMosaicHome())) {
+    throw new Error(
+      `install-systemd only supports the default Mosaic home (${defaultMosaicHome()}) because the user systemd units use %h/.config/mosaic paths.`,
+    );
+  }
+}
+
+export async function loadFleetRoster(path: string): Promise<FleetRoster> {
+  const rawText = await readFile(path, 'utf8');
+  const parsed = parseRosterText(rawText, path);
+  return normalizeRoster(parsed);
+}
+
+export function getRosterAgent(roster: FleetRoster, name: string): FleetAgent {
+  const agent = roster.agents.find((candidate) => candidate.name === name);
+  if (!agent) {
+    throw new Error(`Agent "${name}" is not in the fleet roster.`);
+  }
+  return agent;
+}
+
+export function generateAgentEnv(roster: FleetRoster, agent: FleetAgent): string {
+  const workingDirectory = agent.workingDirectory ?? roster.defaults.workingDirectory;
+  return [
+    `MOSAIC_AGENT_NAME=${shellEnvValue(agent.name)}`,
+    `MOSAIC_AGENT_RUNTIME=${shellEnvValue(agent.runtime)}`,
+    `MOSAIC_AGENT_WORKDIR=${shellEnvValue(expandHome(workingDirectory))}`,
+    `MOSAIC_TMUX_SOCKET=${shellEnvValue(roster.tmux.socketName)}`,
+    '',
+  ].join('\n');
+}
+
+export function buildFleetServiceCommand(action: FleetServiceAction, agentName?: string): string[] {
+  const service = agentName ? `mosaic-agent@${agentName}.service` : 'mosaic-tmux-holder.service';
+  return ['systemctl', '--user', action, service];
+}
+
+export function buildAgentSendCommand(
+  paths: FleetPaths,
+  agentName: string,
+  message: string,
+  socketName = DEFAULT_SOCKET_NAME,
+): string[] {
+  return [
+    join(paths.tmuxToolsDir, 'agent-send.sh'),
+    '-L',
+    socketName,
+    '-s',
+    agentName,
+    '-m',
+    message,
+  ];
+}
+
+export function buildAgentResetCommand(
+  paths: FleetPaths,
+  agentName: string,
+  resetCommand: string,
+  socketName = DEFAULT_SOCKET_NAME,
+): string[] {
+  return [
+    join(paths.tmuxToolsDir, 'send-message.sh'),
+    '-L',
+    socketName,
+    '-t',
+    `=${agentName}`,
+    '-m',
+    resetCommand,
+  ];
+}
+
+export function buildAgentTailCommand(
+  agentName: string,
+  lines: number,
+  socketName = DEFAULT_SOCKET_NAME,
+): string[] {
+  return [
+    'tmux',
+    '-L',
+    socketName,
+    'capture-pane',
+    '-t',
+    `=${agentName}:0.0`,
+    '-p',
+    '-S',
+    `-${lines}`,
+  ];
+}
+
+export function registerFleetCommand(program: Command, deps: FleetCommandDeps = {}): Command {
+  const runner = deps.runner ?? runCommand;
+  const paths = resolveFleetPaths(deps.mosaicHome);
+  const frameworkRoot = deps.frameworkRoot ?? resolveFrameworkRoot();
+
+  const cmd = program
+    .command('fleet')
+    .description('Manage the local Mosaic tmux fleet canary')
+    .option('--mosaic-home <path>', 'Mosaic home directory', paths.mosaicHome)
+    .option('--roster <path>', 'Fleet roster path');
+
+  cmd
+    .command('init')
+    .description('Initialize a local fleet roster')
+    .option('--profile <name>', 'Roster profile: minimal or local-canary', 'minimal')
+    .option('--write', 'Write the roster to Mosaic home')
+    .option('--force', 'Overwrite an existing roster when used with --write')
+    .action(async (opts: { profile: string; write?: boolean; force?: boolean }) => {
+      const commandOpts = cmd.opts<{ mosaicHome: string; roster?: string }>();
+      const activePaths = resolveFleetPaths(commandOpts.mosaicHome);
+      const profile = parseInitProfile(opts.profile);
+      const source = join(frameworkRoot, 'fleet', 'examples', `${profile}.yaml`);
+      const content = await readFile(source, 'utf8');
+      if (!opts.write) {
+        console.log(content.trimEnd());
+        return;
+      }
+      const destination = commandOpts.roster ?? activePaths.rosterPath;
+      if (!opts.force && (await canRead(destination))) {
+        throw new Error(
+          `Fleet roster already exists: ${destination}. Re-run with --force to overwrite.`,
+        );
+      }
+      await mkdir(dirname(destination), { recursive: true });
+      await writeFile(destination, content);
+      console.log(`Wrote fleet roster: ${destination}`);
+    });
+
+  cmd
+    .command('install')
+    .description('Install local fleet tools and user systemd units')
+    .action(async () => installFleet(cmd, frameworkRoot));
+
+  cmd
+    .command('install-systemd')
+    .description('Install local fleet tools and user systemd units')
+    .action(async () => installFleet(cmd, frameworkRoot));
+
+  for (const action of ['start', 'stop', 'restart'] as const) {
+    cmd
+      .command(`${action} [agent]`)
+      .description(`${action} the fleet holder or one agent`)
+      .action(async (agent?: string) => {
+        const roster = await loadRosterForCommand(cmd);
+        if (agent) {
+          getRosterAgent(roster, agent);
+          await runChecked(runner, buildFleetServiceCommand(action, agent));
+          return;
+        }
+        if (action === 'stop') {
+          await stopFleetBestEffort(
+            runner,
+            roster.agents.map((rosterAgent) => rosterAgent.name),
+          );
+          return;
+        }
+        await runChecked(runner, buildFleetServiceCommand(action));
+        for (const rosterAgent of roster.agents) {
+          await runChecked(runner, buildFleetServiceCommand(action, rosterAgent.name));
+        }
+      });
+  }
+
+  cmd
+    .command('status [agent]')
+    .description('Show fleet holder or agent systemd status')
+    .option('--json', 'Print JSON status')
+    .action(async (agent: string | undefined, opts: { json?: boolean }) => {
+      if (agent) {
+        const roster = await loadRosterForCommand(cmd);
+        getRosterAgent(roster, agent);
+      }
+      const result = await runner(...splitCommand(buildFleetServiceCommand('status', agent)));
+      if (opts.json) {
+        console.log(
+          JSON.stringify({
+            exitCode: result.exitCode,
+            stdout: result.stdout,
+            stderr: result.stderr,
+          }),
+        );
+        setExitCodeFromResult(result);
+        return;
+      }
+      writeCommandOutput(result);
+    });
+
+  cmd
+    .command('verify')
+    .description('Verify the local canary holder and roster sessions on the isolated socket')
+    .action(async () => {
+      const roster = await loadRosterForCommand(cmd);
+      const socketName = roster.tmux.socketName;
+      await runChecked(runner, [
+        'tmux',
+        '-L',
+        socketName,
+        'has-session',
+        '-t',
+        `=${roster.tmux.holderSession}:0.0`,
+      ]);
+      for (const agent of roster.agents) {
+        await runChecked(runner, [
+          'tmux',
+          '-L',
+          socketName,
+          'has-session',
+          '-t',
+          `=${agent.name}:0.0`,
+        ]);
+      }
+      console.log(`Verified fleet on tmux socket ${socketName}.`);
+    });
+
+  return cmd;
+}
+
+export function registerFleetAgentCommands(
+  agentCommand: Command,
+  deps: FleetCommandDeps = {},
+): void {
+  const runner = deps.runner ?? runCommand;
+
+  agentCommand
+    .command('roster')
+    .description('List agents from the local fleet roster')
+    .option('--json', 'Print JSON')
+    .action(async (opts: { json?: boolean }) => {
+      const roster = await loadRosterFromAgentCommand(agentCommand, deps.mosaicHome);
+      if (opts.json) {
+        console.log(JSON.stringify(roster, null, 2));
+        return;
+      }
+      for (const agent of roster.agents) {
+        console.log(`${agent.name}\t${agent.runtime}\t${agent.className}`);
+      }
+    });
+
+  agentCommand
+    .command('status [agent]')
+    .description('Show tmux status for the local fleet or one agent')
+    .option('--json', 'Print JSON')
+    .action(async (agent: string | undefined, opts: { json?: boolean }) => {
+      const roster = await loadRosterFromAgentCommand(agentCommand, deps.mosaicHome);
+      if (agent) {
+        getRosterAgent(roster, agent);
+      }
+      const command = agent
+        ? ['tmux', '-L', roster.tmux.socketName, 'has-session', '-t', `=${agent}:0.0`]
+        : ['tmux', '-L', roster.tmux.socketName, 'ls'];
+      const result = await runner(...splitCommand(command));
+      if (opts.json) {
+        console.log(
+          JSON.stringify({
+            exitCode: result.exitCode,
+            stdout: result.stdout,
+            stderr: result.stderr,
+          }),
+        );
+        setExitCodeFromResult(result);
+        return;
+      }
+      writeCommandOutput(result);
+    });
+
+  agentCommand
+    .command('send <agent>')
+    .description('Send a message to a local fleet agent')
+    .requiredOption('--message <text>', 'Message text')
+    .action(async (agent: string, opts: { message: string }) => {
+      const roster = await loadRosterFromAgentCommand(agentCommand, deps.mosaicHome);
+      getRosterAgent(roster, agent);
+      const paths = resolveFleetPaths(resolveMosaicHomeFromCommand(agentCommand, deps.mosaicHome));
+      await runChecked(
+        runner,
+        buildAgentSendCommand(paths, agent, opts.message, roster.tmux.socketName),
+      );
+    });
+
+  agentCommand
+    .command('reset <agent>')
+    .description('Reset a local fleet agent by sending the runtime reset command')
+    .option('--clear', 'Send /clear')
+    .option('--new', 'Send /new')
+    .action(async (agent: string, opts: { clear?: boolean; new?: boolean }) => {
+      const roster = await loadRosterFromAgentCommand(agentCommand, deps.mosaicHome);
+      const rosterAgent = getRosterAgent(roster, agent);
+      const paths = resolveFleetPaths(resolveMosaicHomeFromCommand(agentCommand, deps.mosaicHome));
+      const resetCommand = opts.clear
+        ? '/clear'
+        : opts.new
+          ? '/new'
+          : (roster.runtimes[rosterAgent.runtime]?.resetCommand ?? '/clear');
+      await runChecked(
+        runner,
+        buildAgentResetCommand(paths, agent, resetCommand, roster.tmux.socketName),
+      );
+    });
+
+  agentCommand
+    .command('tail <agent>')
+    .description('Print recent pane output for a local fleet agent')
+    .option('-n, --lines <number>', 'Number of pane history lines', '80')
+    .action(async (agent: string, opts: { lines: string }) => {
+      const roster = await loadRosterFromAgentCommand(agentCommand, deps.mosaicHome);
+      getRosterAgent(roster, agent);
+      const lines = Number.parseInt(opts.lines, 10);
+      const result = await runner(
+        ...splitCommand(
+          buildAgentTailCommand(agent, Number.isFinite(lines) ? lines : 80, roster.tmux.socketName),
+        ),
+      );
+      writeCommandOutput(result);
+    });
+}
+
+async function installFleet(cmd: Command, frameworkRoot: string): Promise<void> {
+  const activePaths = resolveFleetPaths(cmd.opts<{ mosaicHome: string }>().mosaicHome);
+  assertDefaultMosaicHomeForSystemd(activePaths.mosaicHome);
+  const roster = await loadRosterForCommand(cmd);
+  await mkdir(activePaths.fleetToolsDir, { recursive: true });
+  await mkdir(activePaths.tmuxToolsDir, { recursive: true });
+  await mkdir(activePaths.systemdUserDir, { recursive: true });
+  await mkdir(activePaths.agentEnvDir, { recursive: true });
+
+  await copyFile(
+    join(frameworkRoot, 'tools', 'fleet', 'start-agent-session.sh'),
+    join(activePaths.fleetToolsDir, 'start-agent-session.sh'),
+  );
+  await copyFile(
+    join(frameworkRoot, 'tools', 'tmux', 'send-message.sh'),
+    join(activePaths.tmuxToolsDir, 'send-message.sh'),
+  );
+  await copyFile(
+    join(frameworkRoot, 'tools', 'tmux', 'agent-send.sh'),
+    join(activePaths.tmuxToolsDir, 'agent-send.sh'),
+  );
+  await copyFile(
+    join(frameworkRoot, 'systemd', 'user', 'mosaic-tmux-holder.service'),
+    join(activePaths.systemdUserDir, 'mosaic-tmux-holder.service'),
+  );
+  await copyFile(
+    join(frameworkRoot, 'systemd', 'user', 'mosaic-agent@.service'),
+    join(activePaths.systemdUserDir, 'mosaic-agent@.service'),
+  );
+
+  for (const agent of roster.agents) {
+    await writeFile(
+      join(activePaths.agentEnvDir, `${agent.name}.env`),
+      generateAgentEnv(roster, agent),
+    );
+  }
+
+  console.log(`Installed fleet files for ${roster.agents.length} agent(s).`);
+}
+
+async function loadRosterForCommand(cmd: Command): Promise<FleetRoster> {
+  const opts = cmd.opts<{ mosaicHome: string; roster?: string }>();
+  return loadFleetRoster(await resolveRosterPath(opts.mosaicHome, opts.roster));
+}
+
+async function loadRosterFromAgentCommand(
+  command: Command,
+  mosaicHomeOverride?: string,
+): Promise<FleetRoster> {
+  const opts = command.optsWithGlobals<{ mosaicHome?: string; roster?: string }>();
+  const mosaicHome = opts.mosaicHome ?? mosaicHomeOverride ?? defaultMosaicHome();
+  return loadFleetRoster(await resolveRosterPath(mosaicHome, opts.roster));
+}
+
+function resolveMosaicHomeFromCommand(command: Command, override?: string): string {
+  const opts = command.optsWithGlobals<{ mosaicHome?: string }>();
+  return opts.mosaicHome ?? override ?? defaultMosaicHome();
+}
+
+function parseRosterText(text: string, path: string): RawFleetRoster {
+  const trimmed = text.trim();
+  if (path.endsWith('.json')) {
+    return JSON.parse(trimmed) as RawFleetRoster;
+  }
+  return YAML.parse(trimmed) as RawFleetRoster;
+}
+
+function normalizeRoster(raw: RawFleetRoster): FleetRoster {
+  assertObject(raw, 'Fleet roster');
+  assertKnownKeys(raw, 'Fleet roster', [
+    'version',
+    'transport',
+    'tmux',
+    'defaults',
+    'runtimes',
+    'agents',
+  ]);
+  if (raw.tmux !== undefined) {
+    assertObject(raw.tmux, 'Fleet roster tmux');
+    assertKnownKeys(raw.tmux, 'Fleet roster tmux', [
+      'socket_name',
+      'socketName',
+      'holder_session',
+      'holderSession',
+    ]);
+  }
+  if (raw.defaults !== undefined) {
+    assertObject(raw.defaults, 'Fleet roster defaults');
+    assertKnownKeys(raw.defaults, 'Fleet roster defaults', [
+      'working_directory',
+      'workingDirectory',
+    ]);
+  }
+  if (raw.runtimes !== undefined) {
+    assertObject(raw.runtimes, 'Fleet roster runtimes');
+    for (const [runtime, config] of Object.entries(raw.runtimes)) {
+      assertObject(config, `Fleet roster runtime "${runtime}"`);
+      assertKnownKeys(config, `Fleet roster runtime "${runtime}"`, [
+        'reset_command',
+        'resetCommand',
+      ]);
+    }
+  }
+  if (raw.version !== 1) {
+    throw new Error('Fleet roster version must be 1.');
+  }
+  if (raw.transport !== 'tmux') {
+    throw new Error('Fleet roster transport must be "tmux".');
+  }
+  if (!Array.isArray(raw.agents) || raw.agents.length === 0) {
+    throw new Error('Fleet roster must define at least one agent.');
+  }
+
+  const agents = raw.agents.map(normalizeAgent);
+  assertUniqueAgentNames(agents);
+
+  return {
+    version: 1,
+    transport: 'tmux',
+    tmux: {
+      socketName: stringValue(
+        raw.tmux?.socket_name ?? raw.tmux?.socketName,
+        DEFAULT_SOCKET_NAME,
+        'Fleet roster tmux socket_name',
+      ),
+      holderSession: stringValue(
+        raw.tmux?.holder_session ?? raw.tmux?.holderSession,
+        DEFAULT_HOLDER_SESSION,
+        'Fleet roster tmux holder_session',
+      ),
+    },
+    defaults: {
+      workingDirectory: stringValue(
+        raw.defaults?.working_directory ?? raw.defaults?.workingDirectory,
+        DEFAULT_WORKING_DIRECTORY,
+        'Fleet roster defaults working_directory',
+      ),
+    },
+    runtimes: normalizeRuntimes(raw.runtimes as RawFleetRoster['runtimes']),
+    agents,
+  };
+}
+
+function normalizeAgent(raw: NonNullable<RawFleetRoster['agents']>[number]): FleetAgent {
+  assertObject(raw, 'Fleet roster agent');
+  assertKnownKeys(raw, 'Fleet roster agent', [
+    'name',
+    'runtime',
+    'class',
+    'working_directory',
+    'workingDirectory',
+    'model_hint',
+    'modelHint',
+    'persistent_persona',
+    'persistentPersona',
+    'reset_between_tasks',
+    'resetBetweenTasks',
+    'kickstart_template',
+    'kickstartTemplate',
+  ]);
+  const name = stringValue(raw.name, '', 'Fleet roster agent name');
+  const runtime = stringValue(
+    raw.runtime,
+    '',
+    `Fleet roster agent "${name || '<unknown>'}" runtime`,
+  );
+  if (!name || !/^[A-Za-z0-9_.-]+$/.test(name)) {
+    throw new Error(`Invalid fleet agent name: ${name || '<empty>'}`);
+  }
+  if (!runtime) {
+    throw new Error(`Fleet agent "${name}" must define a runtime.`);
+  }
+  return {
+    name,
+    runtime,
+    className: stringValue(raw.class, 'worker', `Fleet roster agent "${name}" class`),
+    workingDirectory: optionalString(
+      raw.working_directory ?? raw.workingDirectory,
+      `Fleet roster agent "${name}" working_directory`,
+    ),
+    modelHint: optionalString(
+      raw.model_hint ?? raw.modelHint,
+      `Fleet roster agent "${name}" model_hint`,
+    ),
+    persistentPersona: optionalBooleanOrString(
+      raw.persistent_persona ?? raw.persistentPersona,
+      `Fleet roster agent "${name}" persistent_persona`,
+    ),
+    resetBetweenTasks: optionalBoolean(
+      raw.reset_between_tasks ?? raw.resetBetweenTasks,
+      `Fleet roster agent "${name}" reset_between_tasks`,
+    ),
+    kickstartTemplate: optionalString(
+      raw.kickstart_template ?? raw.kickstartTemplate,
+      `Fleet roster agent "${name}" kickstart_template`,
+    ),
+  };
+}
+
+function normalizeRuntimes(
+  raw: RawFleetRoster['runtimes'] | undefined,
+): Record<string, { resetCommand: string }> {
+  const result: Record<string, { resetCommand: string }> = { ...DEFAULT_RUNTIME_RESETS };
+  for (const [runtime, config] of Object.entries(raw ?? {})) {
+    result[runtime] = {
+      resetCommand: stringValue(
+        config.reset_command ?? config.resetCommand,
+        '/clear',
+        `Fleet roster runtime "${runtime}" reset_command`,
+      ),
+    };
+  }
+  return result;
+}
+
+function assertObject(value: unknown, label: string): asserts value is Record<string, unknown> {
+  if (!value || typeof value !== 'object' || Array.isArray(value)) {
+    throw new Error(`${label} must be an object.`);
+  }
+}
+
+function assertKnownKeys(
+  value: Record<string, unknown>,
+  label: string,
+  allowedKeys: readonly string[],
+): void {
+  const allowed = new Set(allowedKeys);
+  const unknownKeys = Object.keys(value).filter((key) => !allowed.has(key));
+  if (unknownKeys.length > 0) {
+    throw new Error(`${label} has unknown field(s): ${unknownKeys.join(', ')}.`);
+  }
+}
+
+function assertUniqueAgentNames(agents: FleetAgent[]): void {
+  const seen = new Set<string>();
+  for (const agent of agents) {
+    if (seen.has(agent.name)) {
+      throw new Error(`Fleet roster has duplicate agent name: ${agent.name}.`);
+    }
+    seen.add(agent.name);
+  }
+}
+
+function stringValue(value: unknown, fallback = '', label = 'Value'): string {
+  if (value === undefined) {
+    return fallback;
+  }
+  if (typeof value !== 'string') {
+    throw new Error(`${label} must be a string.`);
+  }
+  return value;
+}
+
+function optionalString(value: unknown, label = 'Value'): string | undefined {
+  if (value === undefined) {
+    return undefined;
+  }
+  if (typeof value !== 'string') {
+    throw new Error(`${label} must be a string.`);
+  }
+  return value;
+}
+
+function optionalBoolean(value: unknown, label = 'Value'): boolean | undefined {
+  if (value === undefined) {
+    return undefined;
+  }
+  if (typeof value !== 'boolean') {
+    throw new Error(`${label} must be a boolean.`);
+  }
+  return value;
+}
+
+function optionalBooleanOrString(value: unknown, label = 'Value'): boolean | string | undefined {
+  if (value === undefined) {
+    return undefined;
+  }
+  if (typeof value !== 'boolean' && typeof value !== 'string') {
+    throw new Error(`${label} must be a boolean or string.`);
+  }
+  return value;
+}
+
+function expandHome(path: string): string {
+  return path === '~' || path.startsWith('~/') ? join(homedir(), path.slice(2)) : path;
+}
+
+function shellEnvValue(value: string): string {
+  if (/^[A-Za-z0-9_./:=@+-]+$/.test(value)) {
+    return value;
+  }
+  return `'${value.replaceAll("'", "'\"'\"'")}'`;
+}
+
+async function stopFleetBestEffort(runner: CommandRunner, agentNames: string[]): Promise<void> {
+  const failures: string[] = [];
+  for (const agentName of agentNames) {
+    const command = buildFleetServiceCommand('stop', agentName);
+    const result = await runner(...splitCommand(command));
+    writeSuccessfulCommandOutput(result);
+    if (result.exitCode !== 0) {
+      failures.push(result.stderr || result.stdout || `Command failed: ${command.join(' ')}`);
+    }
+  }
+
+  const holderCommand = buildFleetServiceCommand('stop');
+  const holderResult = await runner(...splitCommand(holderCommand));
+  writeSuccessfulCommandOutput(holderResult);
+  if (holderResult.exitCode !== 0) {
+    failures.push(
+      holderResult.stderr || holderResult.stdout || `Command failed: ${holderCommand.join(' ')}`,
+    );
+  }
+
+  if (failures.length > 0) {
+    throw new Error(
+      `Fleet stop completed with ${failures.length} failure(s): ${failures.join('; ')}`,
+    );
+  }
+}
+
+async function runChecked(runner: CommandRunner, command: string[]): Promise<void> {
+  const result = await runner(...splitCommand(command));
+  if (result.exitCode !== 0) {
+    throw new Error(result.stderr || result.stdout || `Command failed: ${command.join(' ')}`);
+  }
+  if (result.stdout) {
+    process.stdout.write(result.stdout);
+  }
+}
+
+function splitCommand(command: string[]): [string, string[]] {
+  const [bin, ...args] = command;
+  if (!bin) {
+    throw new Error('Cannot run an empty command.');
+  }
+  return [bin, args];
+}
+
+function parseInitProfile(profile: string): 'minimal' | 'local-canary' {
+  if (profile === 'minimal' || profile === 'local-canary') {
+    return profile;
+  }
+  throw new Error(`Unsupported fleet profile "${profile}". Use: minimal, local-canary.`);
+}
+
+function writeCommandOutput(result: CommandResult): void {
+  if (result.stdout) {
+    process.stdout.write(result.stdout);
+  } else if (result.stderr) {
+    process.stderr.write(result.stderr);
+  }
+  setExitCodeFromResult(result);
+}
+
+function writeSuccessfulCommandOutput(result: CommandResult): void {
+  if (result.exitCode !== 0) {
+    return;
+  }
+  if (result.stdout) {
+    process.stdout.write(result.stdout);
+  }
+}
+
+function setExitCodeFromResult(result: CommandResult): void {
+  if (result.exitCode !== 0) {
+    process.exitCode = result.exitCode;
+  }
+}
+
+function runCommand(command: string, args: string[]): Promise<CommandResult> {
+  return new Promise((resolvePromise) => {
+    const child = spawn(command, args, { stdio: ['ignore', 'pipe', 'pipe'] });
+    let stdout = '';
+    let stderr = '';
+    child.stdout.on('data', (chunk: Buffer) => {
+      stdout += chunk.toString('utf8');
+    });
+    child.stderr.on('data', (chunk: Buffer) => {
+      stderr += chunk.toString('utf8');
+    });
+    child.on('error', (error) => {
+      resolvePromise({ stdout, stderr: error.message, exitCode: 127 });
+    });
+    child.on('close', (code) => {
+      resolvePromise({ stdout, stderr, exitCode: code ?? 1 });
+    });
+  });
+}
+
+function resolveFrameworkRoot(): string {
+  const currentFile = fileURLToPath(import.meta.url);
+  return resolve(dirname(currentFile), '..', '..', 'framework');
+}
+
+async function canRead(path: string): Promise<boolean> {
+  try {
+    await access(path, constants.R_OK);
+    return true;
+  } catch {
+    return false;
+  }
+}
+
+export async function resolveRosterPath(
+  mosaicHome: string,
+  explicitPath?: string,
+): Promise<string> {
+  if (explicitPath) {
+    return explicitPath;
+  }
+  const yamlPath = resolveFleetPaths(mosaicHome).rosterPath;
+  if (await canRead(yamlPath)) {
+    return yamlPath;
+  }
+  const jsonPath = join(mosaicHome, 'fleet', 'roster.json');
+  return jsonPath;
+}

packages/mosaic/src/commands/launch.spec.ts

@@ -1,6 +1,15 @@
 import { describe, it, expect, vi, beforeEach, afterEach, type MockInstance } from 'vitest';
 import { Command } from 'commander';
-import { buildPiSkillArgs, registerRuntimeLaunchers, type RuntimeLaunchHandler } from './launch.js';
+import { mkdtempSync, mkdirSync, writeFileSync, symlinkSync, rmSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import {
+  buildPiSkillArgs,
+  enumerateSkillDirs,
+  piForceSkillNames,
+  registerRuntimeLaunchers,
+  type RuntimeLaunchHandler,
+} from './launch.js';
 
 /**
  * Tests for the commander wiring between `mosaic <runtime>` / `mosaic yolo <runtime>`
@@ -23,6 +32,7 @@ function buildProgram(handler: RuntimeLaunchHandler): Command {
 }
 
 const fakeSkills = ['--skill', '/skills/test-driven-development', '--skill', '/skills/pdf'];
+const fakeForced = ['--skill', '/skills/mosaic-tools'];
 
 // `process.exit` returns `never`, so vi.spyOn demands a replacement with the
 // same signature. We throw from the mock to short-circuit into test-land.
@@ -66,16 +76,42 @@ describe('registerRuntimeLaunchers — non-yolo subcommands', () => {
 });
 
 describe('buildPiSkillArgs', () => {
-  it('defaults to disabling Pi skill discovery to keep startup context small', () => {
-    expect(buildPiSkillArgs([], {}, fakeSkills)).toEqual(['--no-skills']);
+  it('disables auto-discovery but force-loads fleet-critical skills by default', () => {
+    expect(buildPiSkillArgs([], {}, fakeSkills, fakeForced)).toEqual([
+      '--no-skills',
+      '--skill',
+      '/skills/mosaic-tools',
+    ]);
   });
 
-  it('keeps explicit user skills while disabling automatic discovery', () => {
-    expect(buildPiSkillArgs(['--skill', '/tmp/custom'], {}, fakeSkills)).toEqual(['--no-skills']);
+  it('ignores _runtimeArgs (user --skill flags reach Pi via the launch handler, not here)', () => {
+    expect(buildPiSkillArgs(['--skill', '/tmp/custom'], {}, fakeSkills, fakeForced)).toEqual([
+      '--no-skills',
+      '--skill',
+      '/skills/mosaic-tools',
+    ]);
   });
 
-  it('supports legacy all-skills mode without double-loading settings skills', () => {
-    expect(buildPiSkillArgs([], { MOSAIC_PI_SKILL_MODE: 'all' }, fakeSkills)).toEqual([
+  it('emits only --no-skills when no forced skills are present on disk', () => {
+    expect(buildPiSkillArgs([], {}, fakeSkills, [])).toEqual(['--no-skills']);
+  });
+
+  it('all-skills mode merges the forced set in without duplicating discovered skills', () => {
+    expect(buildPiSkillArgs([], { MOSAIC_PI_SKILL_MODE: 'all' }, fakeSkills, fakeForced)).toEqual([
+      '--no-skills',
+      '--skill',
+      '/skills/test-driven-development',
+      '--skill',
+      '/skills/pdf',
+      '--skill',
+      '/skills/mosaic-tools',
+    ]);
+  });
+
+  it('all-skills mode does not double-load a forced skill already discovered', () => {
+    expect(
+      buildPiSkillArgs([], { MOSAIC_PI_SKILL_MODE: 'all' }, fakeSkills, ['--skill', '/skills/pdf']),
+    ).toEqual([
       '--no-skills',
       '--skill',
       '/skills/test-driven-development',
@@ -84,8 +120,117 @@ describe('buildPiSkillArgs', () => {
     ]);
   });
 
-  it('supports native Pi discovery when explicitly requested', () => {
-    expect(buildPiSkillArgs([], { MOSAIC_PI_SKILL_MODE: 'discover' }, fakeSkills)).toEqual([]);
+  it('force-loads fleet skills under native Pi discovery when not already discoverable', () => {
+    // Empty native set => Pi would not find mosaic-tools on its own, so force it.
+    expect(
+      buildPiSkillArgs([], { MOSAIC_PI_SKILL_MODE: 'discover' }, fakeSkills, fakeForced, new Set()),
+    ).toEqual(['--skill', '/skills/mosaic-tools']);
+  });
+
+  it('discover mode drops a forced skill Pi already discovers natively (no double-load)', () => {
+    // mosaic-tools is reachable from a Pi native root, so native discovery
+    // covers it — forcing it again would register the same skill twice.
+    expect(
+      buildPiSkillArgs(
+        [],
+        { MOSAIC_PI_SKILL_MODE: 'discover' },
+        fakeSkills,
+        fakeForced,
+        new Set(['/skills/mosaic-tools']),
+      ),
+    ).toEqual([]);
+  });
+
+  it('discover mode keeps a forced skill that no native root provides', () => {
+    expect(
+      buildPiSkillArgs(
+        [],
+        { MOSAIC_PI_SKILL_MODE: 'discover' },
+        fakeSkills,
+        fakeForced,
+        new Set(['/skills/some-other-skill']),
+      ),
+    ).toEqual(['--skill', '/skills/mosaic-tools']);
+  });
+
+  it('discover mode collapses a forced skill listed twice to a single --skill', () => {
+    // Mirror 'all' mode: intra-forced-set duplicates (same realpath) dedup.
+    expect(
+      buildPiSkillArgs(
+        [],
+        { MOSAIC_PI_SKILL_MODE: 'discover' },
+        fakeSkills,
+        ['--skill', '/skills/mosaic-tools', '--skill', '/skills/mosaic-tools'],
+        new Set(),
+      ),
+    ).toEqual(['--skill', '/skills/mosaic-tools']);
+  });
+});
+
+describe('enumerateSkillDirs (real FS)', () => {
+  let root: string;
+
+  beforeEach(() => {
+    root = mkdtempSync(join(tmpdir(), 'mosaic-skills-'));
+  });
+
+  afterEach(() => {
+    rmSync(root, { recursive: true, force: true });
+  });
+
+  function makeSkill(parent: string, name: string): string {
+    const dir = join(parent, name);
+    mkdirSync(dir, { recursive: true });
+    writeFileSync(join(dir, 'SKILL.md'), `# ${name}\n`);
+    return dir;
+  }
+
+  it('accepts a symlinked skill dir (regression: synced fleet skills are symlinks)', () => {
+    // Real skill lives under `canonical/`; the scanned root only has a symlink to it.
+    const canonical = makeSkill(join(root, 'canonical'), 'mosaic-tools');
+    const scanned = join(root, 'scanned');
+    mkdirSync(scanned, { recursive: true });
+    symlinkSync(canonical, join(scanned, 'mosaic-tools'), 'dir');
+
+    expect(enumerateSkillDirs([scanned])).toEqual(['--skill', join(scanned, 'mosaic-tools')]);
+  });
+
+  it('dedups by real path when the same skill is reachable from two roots', () => {
+    // Root A holds the real dir; root B symlinks to it — one --skill, not two.
+    const rootA = join(root, 'a');
+    const rootB = join(root, 'b');
+    const real = makeSkill(rootA, 'mosaic-tools');
+    mkdirSync(rootB, { recursive: true });
+    symlinkSync(real, join(rootB, 'mosaic-tools'), 'dir');
+
+    expect(enumerateSkillDirs([rootA, rootB])).toEqual(['--skill', real]);
+  });
+
+  it('skips directories without a SKILL.md and missing roots', () => {
+    mkdirSync(join(root, 'present', 'not-a-skill'), { recursive: true });
+    makeSkill(join(root, 'present'), 'real-skill');
+
+    expect(enumerateSkillDirs([join(root, 'present'), join(root, 'does-not-exist')])).toEqual([
+      '--skill',
+      join(root, 'present', 'real-skill'),
+    ]);
+  });
+});
+
+describe('piForceSkillNames', () => {
+  it('defaults to mosaic-tools when MOSAIC_PI_FORCE_SKILLS is unset', () => {
+    expect(piForceSkillNames({})).toEqual(['mosaic-tools']);
+  });
+
+  it('treats an empty string as "disable force-loading" (distinct from unset)', () => {
+    expect(piForceSkillNames({ MOSAIC_PI_FORCE_SKILLS: '' })).toEqual([]);
+  });
+
+  it('parses a colon list, trimming blanks and whitespace', () => {
+    expect(piForceSkillNames({ MOSAIC_PI_FORCE_SKILLS: 'mosaic-tools: mosaic-gitea ::' })).toEqual([
+      'mosaic-tools',
+      'mosaic-gitea',
+    ]);
   });
 });
 

packages/mosaic/src/commands/launch.ts

@@ -6,7 +6,15 @@
  */
 
 import { execFileSync, execSync, spawnSync } from 'node:child_process';
-import { existsSync, mkdirSync, readFileSync, writeFileSync, readdirSync, rmSync } from 'node:fs';
+import {
+  existsSync,
+  mkdirSync,
+  readFileSync,
+  writeFileSync,
+  readdirSync,
+  realpathSync,
+  rmSync,
+} from 'node:fs';
 import { createRequire } from 'node:module';
 import { homedir } from 'node:os';
 import { join, dirname } from 'node:path';
@@ -428,25 +436,74 @@ function ensureRuntimeConfig(runtime: RuntimeName, destPath: string): void {
 
 // ─── Pi skill/extension discovery ────────────────────────────────────────────
 
-function discoverPiSkills(): string[] {
+/** Resolve a skill dir to its canonical real path so symlinked duplicates
+ * (e.g. ~/.pi/agent/skills/X -> ~/.config/mosaic/skills/X) collapse to one key.
+ * Falls back to the literal path if it can't be resolved (e.g. broken link). */
+function skillRealPath(dir: string): string {
+  try {
+    return realpathSync(dir);
+  } catch {
+    return dir;
+  }
+}
+
+/** Skill roots Pi auto-discovers natively (no `--skill` needed): its global
+ * skills dir and the project-local one relative to the launch cwd. */
+function piNativeSkillRoots(cwd: string = process.cwd()): string[] {
+  return [join(homedir(), '.pi', 'agent', 'skills'), join(cwd, '.pi', 'skills')];
+}
+
+/** Enumerate skill dirs under a set of roots, deduped by real path. A directory
+ * counts as a skill when it (or its symlink target) contains a SKILL.md.
+ * Exported for tests (real-FS coverage of symlink acceptance + realpath dedup). */
+export function enumerateSkillDirs(roots: string[]): string[] {
+  const seen = new Set<string>();
   const args: string[] = [];
-  for (const skillsRoot of [join(MOSAIC_HOME, 'skills'), join(MOSAIC_HOME, 'skills-local')]) {
+  for (const skillsRoot of roots) {
     if (!existsSync(skillsRoot)) continue;
     try {
       for (const entry of readdirSync(skillsRoot, { withFileTypes: true })) {
-        if (!entry.isDirectory()) continue;
+        // Synced fleet skills land as symlinks, so accept both dirs and links.
+        if (!entry.isDirectory() && !entry.isSymbolicLink()) continue;
         const skillDir = join(skillsRoot, entry.name);
-        if (existsSync(join(skillDir, 'SKILL.md'))) {
-          args.push('--skill', skillDir);
-        }
+        if (!existsSync(join(skillDir, 'SKILL.md'))) continue;
+        const key = skillRealPath(skillDir);
+        if (seen.has(key)) continue;
+        seen.add(key);
+        args.push('--skill', skillDir);
       }
     } catch {
-      // skip
+      // skip unreadable roots
     }
   }
   return args;
 }
 
+/** Every skill dir Pi would link under `MOSAIC_PI_SKILL_MODE=all`: the Mosaic
+ * global/local catalog plus Pi's own native roots. `--no-skills` suppresses
+ * native auto-discovery, so 'all' must re-add the native roots explicitly or
+ * they would be silently dropped. Deduped by real path. */
+function discoverPiSkills(cwd: string = process.cwd()): string[] {
+  return enumerateSkillDirs([
+    join(MOSAIC_HOME, 'skills'),
+    join(MOSAIC_HOME, 'skills-local'),
+    ...piNativeSkillRoots(cwd),
+  ]);
+}
+
+/** Real paths of skills Pi will auto-discover from its native roots. Used to
+ * drop redundant force-loads in 'discover' mode (which keeps native discovery
+ * on) so the same skill is not registered twice. */
+function piNativeSkillRealPaths(cwd: string = process.cwd()): Set<string> {
+  const args = enumerateSkillDirs(piNativeSkillRoots(cwd));
+  const set = new Set<string>();
+  for (let i = 1; i < args.length; i += 2) {
+    const dir = args[i];
+    if (dir !== undefined) set.add(skillRealPath(dir));
+  }
+  return set;
+}
+
 type PiSkillMode = 'none' | 'all' | 'discover';
 
 function normalizePiSkillMode(env: NodeJS.ProcessEnv): PiSkillMode {
@@ -455,22 +512,96 @@ function normalizePiSkillMode(env: NodeJS.ProcessEnv): PiSkillMode {
   return 'none';
 }
 
+/**
+ * Fleet-critical Pi skills that are force-loaded on every Pi launch regardless
+ * of MOSAIC_PI_SKILL_MODE. They cover the highest-frequency cross-agent and
+ * git-provider operations where Pi workers historically improvised raw CLIs
+ * (raw `tmux send-keys`, raw `tea`/`gh`/`glab`) instead of the maintained
+ * `~/.config/mosaic/tools/` wrappers.
+ *
+ * An explicit `--skill <dir>` overrides `--no-skills` for that path, so forcing
+ * a single targeted skill surfaces the must-use toolkit without loading the full
+ * ~100-skill catalog (context bloat). Missing skills are skipped silently, so
+ * this is a no-op until the named skill is synced into ~/.config/mosaic/skills/.
+ *
+ * Override with MOSAIC_PI_FORCE_SKILLS (colon-separated skill dir names; set to
+ * an empty string to disable force-loading entirely).
+ */
+const DEFAULT_PI_FORCE_SKILLS = ['mosaic-tools'];
+
+export function piForceSkillNames(env: NodeJS.ProcessEnv): string[] {
+  const override = env['MOSAIC_PI_FORCE_SKILLS'];
+  if (override === undefined) return DEFAULT_PI_FORCE_SKILLS;
+  return override
+    .split(':')
+    .map((name) => name.trim())
+    .filter(Boolean);
+}
+
+function forcedPiSkillArgs(env: NodeJS.ProcessEnv = process.env): string[] {
+  const args: string[] = [];
+  for (const name of piForceSkillNames(env)) {
+    const skillDir = join(MOSAIC_HOME, 'skills', name);
+    if (existsSync(join(skillDir, 'SKILL.md'))) {
+      args.push('--skill', skillDir);
+    }
+  }
+  return args;
+}
+
+/** Concatenate `--skill <dir>` arg groups, dropping any skill already seen.
+ * Dedup is by real path, so a forced skill and the same skill reached via a
+ * different (e.g. symlinked) directory collapse to a single `--skill`. */
+function mergeSkillArgs(...groups: string[][]): string[] {
+  const seen = new Set<string>();
+  const out: string[] = [];
+  for (const group of groups) {
+    for (let i = 0; i < group.length; i += 2) {
+      const dir = group[i + 1];
+      if (group[i] !== '--skill' || dir === undefined) continue;
+      const key = skillRealPath(dir);
+      if (seen.has(key)) continue;
+      seen.add(key);
+      out.push('--skill', dir);
+    }
+  }
+  return out;
+}
+
 export function buildPiSkillArgs(
   _runtimeArgs: string[],
   env: NodeJS.ProcessEnv = process.env,
   discoveredSkillArgs: string[] = discoverPiSkills(),
+  forcedSkillArgs: string[] = forcedPiSkillArgs(env),
+  nativeSkillRealPaths: Set<string> = piNativeSkillRealPaths(),
 ): string[] {
   const mode = normalizePiSkillMode(env);
 
   if (mode === 'discover') {
-    return [];
+    // Native Pi discovery stays on, so only force-load fleet skills it will NOT
+    // already find under its native roots — otherwise the same skill is
+    // registered twice (once natively, once via --skill). mergeSkillArgs first
+    // collapses any intra-forced-set realpath duplicates, mirroring 'all' mode.
+    const deduped = mergeSkillArgs(forcedSkillArgs);
+    const out: string[] = [];
+    for (let i = 0; i < deduped.length; i += 2) {
+      const dir = deduped[i + 1];
+      if (deduped[i] !== '--skill' || dir === undefined) continue;
+      if (nativeSkillRealPaths.has(skillRealPath(dir))) continue;
+      out.push('--skill', dir);
+    }
+    return out;
   }
 
   if (mode === 'all') {
-    return ['--no-skills', ...discoveredSkillArgs];
+    // 'all' links the full catalog; merge in the forced set so fleet-critical
+    // skills are guaranteed present even if they live only under skills-local/.
+    // discoverPiSkills already covers Pi's native roots, which `--no-skills`
+    // would otherwise suppress.
+    return ['--no-skills', ...mergeSkillArgs(discoveredSkillArgs, forcedSkillArgs)];
   }
 
-  return ['--no-skills'];
+  return ['--no-skills', ...forcedSkillArgs];
 }
 
 function discoverPiExtension(): string[] {

scratchpads/2026-06-19-tmux-fleet-durable-install-plan.md

@@ -0,0 +1,755 @@
+# Durable tmux Fleet Installation Plan
+
+> **For Mosaic/Hermes:** This is an implementation plan for making the tmux-backed Mosaic software-factory fleet durable on this server and reusable in generic Mosaic Stack installs. Keep local USC/Mosaic defaults in profiles; keep framework behavior customizable.
+
+**Goal:** Add a supported Mosaic tmux-fleet installation path: holder-owned tmux server, per-agent reusable sessions, reliable send/reset/status tools, local roster customization, and a documented cutover for this server.
+
+**Architecture:** Mosaic should ship generic tmux fleet primitives in the framework, then layer local rosters through configuration. The holder service owns the tmux socket; each agent service joins the holder-owned server and runs `mosaic yolo <runtime>`. The orchestrator addresses agents through `mosaic agent ...` abstractions so tmux can later be replaced by Matrix-backed agent comms without changing mission flow.
+
+**Reference:** AI Guide `playbooks/tmux-fleet.md` at commit `2a0b0b5` documents the organization-neutral holder-service pattern, exact-match `=<name>` stop targets, and coupled-server cutover/verification sequence. The Stack implementation should treat that as the lifecycle model and keep concrete Mosaic unit/tooling details here.
+
+**Tech Stack:** Bash, tmux, user systemd units, Mosaic CLI/framework installer, JSON/YAML roster config, existing `packages/mosaic/framework/tools/tmux/{agent-send.sh,send-message.sh}`.
+
+---
+
+## Current evidence from this server
+
+Checked 2026-06-19:
+
+- Host: `W-jarvis`
+- User: `jarvis`
+- tmux: `/usr/bin/tmux`, version `3.4`
+- user systemd: active
+- existing tmux sessions: `ai-bma-0`, `dyor-1`, `melaniewoltje-3`, `sage-2`
+- existing Mosaic runtime: `/home/jarvis/.npm-global/bin/mosaic`, version `0.0.31`
+- installed `~/.config/mosaic/tools/tmux` was not present even though the stack repo contains `packages/mosaic/framework/tools/tmux/`
+
+Implication: do not kill the current tmux server casually. This server has active ad-hoc/service sessions. The durable fleet cutover must be planned, with either a separate socket first or a scheduled fleet recycle.
+
+## Design decisions
+
+### 1. Generic framework, local profile
+
+The Mosaic framework should ship:
+
+- systemd unit templates;
+- tmux fleet CLI wrappers;
+- roster schema and examples;
+- install/enable/status/reset commands;
+- docs and verification scripts.
+
+Local environments should provide:
+
+- agent names;
+- runtime per slot (`claude`, `pi`, `codex`, etc.);
+- default role class;
+- launch directory;
+- optional kickstart prompt;
+- model/provider hints;
+- transport selection (`tmux` now, `matrix` later).
+
+Do not bake the USC roster into generic install code. Ship it as an example profile.
+
+### 2. Durable sessions, disposable task context
+
+Session names are durable operational addresses. Task persona is disposable. Reusable worker slots should be reset with `/clear` or `/new` and then receive a fresh task kickstart.
+
+Persistent/semi-persistent personas:
+
+- lead orchestrator;
+- final/adversarial reviewer;
+- architecture/enhancement lane.
+
+Disposable slots:
+
+- implementers;
+- ordinary reviewers;
+- security reviewers unless actively holding a security mission.
+
+### 3. Transport abstraction now
+
+Add commands around tmux instead of calling tmux directly from orchestration:
+
+```bash
+mosaic agent send <agent> --message "..."
+mosaic agent status [--json]
+mosaic agent reset <agent> [--clear|--new]
+mosaic agent roster [--json]
+mosaic fleet install|start|stop|restart|status|verify
+```
+
+Today these call tmux/systemd. Later the same command surface can target Matrix or per-agent gateways.
+
+### 4. Avoid shared-server ownership bug
+
+Use the AI Guide holder pattern:
+
+```text
+mosaic-tmux-holder.service owns the tmux server/socket
+mosaic-agent@<name>.service joins the existing holder-owned socket
+ExecStop kills only session =<name>
+```
+
+Use exact tmux targets: `=<session>`.
+
+### 5. Prefer separate named socket for Mosaic factory
+
+To avoid disturbing existing tmux work, the default fleet should use a named socket such as:
+
+```text
+$XDG_RUNTIME_DIR/mosaic-factory.tmux
+```
+
+or tmux socket name:
+
+```bash
+tmux -L mosaic-factory ...
+```
+
+This avoids collision with ordinary `tmux ls` sessions. The send tools need socket support.
+
+---
+
+## Target USC-style roster example
+
+Ship as example only, not default:
+
+```yaml
+version: 1
+transport: tmux
+tmux:
+  socket_name: mosaic-factory
+  holder_session: _holder
+  working_directory: ~/src
+agents:
+  - name: mos-claude
+    runtime: claude
+    class: orchestrator
+    model_hint: Claude Opus
+    persistent_persona: true
+  - name: coder0
+    runtime: claude
+    class: implementer
+    model_hint: Claude Opus
+    reset_between_tasks: true
+  - name: coder1
+    runtime: claude
+    class: implementer
+    model_hint: Claude Opus
+    reset_between_tasks: true
+  - name: coder2
+    runtime: pi
+    class: implementer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: coder3
+    runtime: pi
+    class: implementer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: coder4
+    runtime: claude
+    class: implementer
+    model_hint: Claude Opus
+    reset_between_tasks: true
+  - name: coder5
+    runtime: claude
+    class: implementer
+    model_hint: Claude Opus
+    reset_between_tasks: true
+  - name: enhance
+    runtime: claude
+    class: enhancer
+    model_hint: Claude Opus
+    persistent_persona: semi
+  - name: rev0
+    runtime: pi
+    class: reviewer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: rev1
+    runtime: pi
+    class: reviewer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: secrev0
+    runtime: pi
+    class: security_reviewer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: secrev1
+    runtime: pi
+    class: security_reviewer
+    model_hint: Pi GPT-5.5
+    reset_between_tasks: true
+  - name: ultron
+    runtime: pi
+    class: final_reviewer
+    model_hint: Pi GPT-5.5
+    persistent_persona: semi
+```
+
+---
+
+## Phase 0 — Confirm install surfaces
+
+### Task 0.1: Inspect installer copy behavior
+
+**Objective:** Confirm how framework files under `packages/mosaic/framework/` become installed under `~/.config/mosaic/`.
+
+**Files:**
+
+- Read: `tools/install.sh`
+- Read: `packages/mosaic/framework/install.sh`
+- Read: `packages/mosaic/src/runtime/install-manifest.ts`
+
+**Steps:**
+
+1. Verify `packages/mosaic/framework/install.sh` rsyncs `tools/tmux`.
+2. Verify whether npm-packaged installs include `framework/tools/tmux`.
+3. Confirm whether installed hosts should run `mosaic update`, `bash tools/install.sh`, or `packages/mosaic/framework/install.sh` to receive new tmux tools.
+4. Record exact propagation command in docs.
+
+**Verification:**
+
+```bash
+bash packages/mosaic/framework/install.sh --help || true
+npm pack --dry-run --json | jq '.[0].files[].path' | grep 'framework/tools/tmux'
+```
+
+Expected: tmux tools are included in installable package or packaging fix is identified.
+
+### Task 0.2: Inspect current yolo launch semantics
+
+**Objective:** Confirm `mosaic yolo claude` and `mosaic yolo pi` accept optional initial prompt text and behave well under systemd/tmux.
+
+**Files:**
+
+- Read: `packages/mosaic/src/**`
+- Read: `packages/mosaic/framework/runtime/claude/RUNTIME.md`
+- Read: `packages/mosaic/framework/runtime/pi/RUNTIME.md`
+
+**Verification commands:**
+
+```bash
+mosaic yolo claude --help
+mosaic yolo pi --help
+```
+
+Expected: a systemd `ExecStart` can launch the runtime either with no prompt or with a kickstart prompt file/string.
+
+---
+
+## Phase 1 — Framework tmux primitives
+
+### Task 1.1: Add socket support to send tools
+
+**Objective:** Allow `agent-send.sh` and `send-message.sh` to target a named Mosaic tmux socket without affecting default tmux sessions.
+
+**Files:**
+
+- Modify: `packages/mosaic/framework/tools/tmux/send-message.sh`
+- Modify: `packages/mosaic/framework/tools/tmux/agent-send.sh`
+- Modify: `packages/mosaic/framework/tools/tmux/README.md`
+- Test: `packages/mosaic/framework/tools/tmux/test-send-message.sh` (new)
+
+**Design:**
+
+Add optional flags:
+
+```bash
+-L SOCKET_NAME      # tmux -L socket name
+-SOCKET PATH        # optional later if needed; avoid conflict with existing -S source label in agent-send
+```
+
+Because `agent-send.sh` already uses `-S` for source label, prefer `-L` for socket name and `-T` or `--socket-path` only if long-option parsing is added.
+
+**Implementation notes:**
+
+- Build a tmux command array:
+
+```bash
+tmux_cmd=(tmux)
+if [ -n "$SOCKET_NAME" ]; then tmux_cmd+=( -L "$SOCKET_NAME" ); fi
+```
+
+- Replace raw `tmux ...` calls with `"${tmux_cmd[@]}" ...`.
+- Pass `-L` through remote ssh invocation.
+- Include socket name in verbose output.
+
+**Verification:**
+
+```bash
+tmux -L mosaic-test new-session -d -s target 'cat'
+packages/mosaic/framework/tools/tmux/send-message.sh -L mosaic-test -t target -m 'hello'
+tmux -L mosaic-test capture-pane -t target -p | grep hello
+tmux -L mosaic-test kill-server
+```
+
+Expected: message lands in the named socket session; default `tmux ls` is untouched.
+
+### Task 1.2: Add exact target validation helper
+
+**Objective:** Prevent accidental prefix targeting in all tmux fleet operations.
+
+**Files:**
+
+- Create: `packages/mosaic/framework/tools/tmux/_lib.sh`
+- Modify: `send-message.sh`
+- Modify: `agent-send.sh`
+
+**Behavior:**
+
+- For session-only agent names, normalize target to `=<name>` before kill/status/reset operations.
+- For explicit pane targets like `session:window.pane`, allow as advanced path but document the risk.
+
+**Verification:**
+
+Create sessions `agent` and `agent0`; verify killing/resetting `agent` does not affect `agent0`.
+
+---
+
+## Phase 2 — systemd unit templates
+
+### Task 2.1: Add holder service template
+
+**Objective:** Ship a user systemd unit template that owns the Mosaic factory tmux server.
+
+**Files:**
+
+- Create: `packages/mosaic/framework/systemd/user/mosaic-tmux-holder.service`
+- Create: `packages/mosaic/framework/tools/fleet/install-user-units.sh`
+
+**Unit shape:**
+
+```ini
+[Unit]
+Description=Mosaic tmux fleet holder
+Documentation=https://git.mosaicstack.dev/mosaicstack/aiguide
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+Environment=MOSAIC_TMUX_SOCKET=mosaic-factory
+ExecStart=/usr/bin/tmux -L ${MOSAIC_TMUX_SOCKET} new-session -d -s _holder 'while true; do sleep 3600; done'
+ExecStop=-/usr/bin/tmux -L ${MOSAIC_TMUX_SOCKET} kill-server
+
+[Install]
+WantedBy=default.target
+```
+
+**Important:** systemd environment expansion in `ExecStart` is limited. Verify syntax; if `%E`/environment expansion is awkward, generate concrete units from config instead of relying on dynamic expansion.
+
+**Verification:**
+
+```bash
+systemd-analyze --user verify ~/.config/systemd/user/mosaic-tmux-holder.service
+systemctl --user daemon-reload
+systemctl --user start mosaic-tmux-holder.service
+tmux -L mosaic-factory ls | grep _holder
+```
+
+### Task 2.2: Add agent service template
+
+**Objective:** Ship a user systemd template that starts one configured agent slot.
+
+**Files:**
+
+- Create: `packages/mosaic/framework/systemd/user/mosaic-agent@.service`
+- Modify: `packages/mosaic/framework/tools/fleet/install-user-units.sh`
+
+**Unit shape:**
+
+```ini
+[Unit]
+Description=Mosaic agent session %i
+Requires=mosaic-tmux-holder.service
+After=mosaic-tmux-holder.service
+PartOf=mosaic-tmux-holder.service
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+WorkingDirectory=%h/src
+Environment=MOSAIC_TMUX_SOCKET=mosaic-factory
+ExecStart=/bin/bash -lc 'tmux -L "$MOSAIC_TMUX_SOCKET" new-session -d -s "%i" "mosaic yolo $(mosaic fleet runtime %i)"'
+ExecStop=-/usr/bin/tmux -L mosaic-factory kill-session -t '=%i'
+
+[Install]
+WantedBy=default.target
+```
+
+**Design warning:** command substitution in unit files can become brittle. Prefer a generated per-agent EnvironmentFile:
+
+```text
+~/.config/mosaic/fleet/agents/coder0.env
+```
+
+with:
+
+```bash
+MOSAIC_AGENT_NAME=coder0
+MOSAIC_AGENT_RUNTIME=claude
+MOSAIC_AGENT_WORKDIR=/home/jarvis/src
+MOSAIC_TMUX_SOCKET=mosaic-factory
+```
+
+Then `ExecStart` calls a wrapper:
+
+```bash
+~/.config/mosaic/tools/fleet/start-agent-session.sh
+```
+
+**Verification:**
+
+```bash
+systemd-analyze --user verify ~/.config/systemd/user/mosaic-agent@.service
+systemctl --user start mosaic-agent@coder0.service
+tmux -L mosaic-factory has-session -t '=coder0'
+systemctl --user restart mosaic-agent@coder0.service
+```
+
+Expected: holder server PID remains unchanged; only `coder0` session recycles.
+
+### Task 2.3: Add start-agent wrapper
+
+**Objective:** Keep systemd units simple by moving config lookup and launch command construction into a script.
+
+**Files:**
+
+- Create: `packages/mosaic/framework/tools/fleet/start-agent-session.sh`
+
+**Behavior:**
+
+Inputs:
+
+```bash
+start-agent-session.sh <agent-name>
+```
+
+Reads:
+
+```text
+$MOSAIC_HOME/fleet/agents/<agent-name>.env
+```
+
+Starts:
+
+```bash
+tmux -L "$MOSAIC_TMUX_SOCKET" new-session -d -s "$MOSAIC_AGENT_NAME" -c "$MOSAIC_AGENT_WORKDIR" "mosaic yolo $MOSAIC_AGENT_RUNTIME"
+```
+
+Guardrails:
+
+- fail if runtime is empty;
+- fail if workdir does not exist;
+- no duplicate sessions unless `--replace` is passed;
+- exact session names only.
+
+---
+
+## Phase 3 — roster config and CLI wrappers
+
+### Task 3.1: Add fleet config schema and examples
+
+**Objective:** Define customizable install-time roster without hardcoding USC.
+
+**Files:**
+
+- Create: `packages/mosaic/framework/fleet/roster.schema.json`
+- Create: `packages/mosaic/framework/fleet/examples/minimal.yaml`
+- Create: `packages/mosaic/framework/fleet/examples/usc-software-factory.yaml`
+- Create: `packages/mosaic/framework/fleet/README.md`
+
+**Schema concepts:**
+
+- `transport`: `tmux` now; `matrix` later.
+- `tmux.socket_name`
+- `tmux.holder_session`
+- `defaults.working_directory`
+- `agents[].name`
+- `agents[].runtime`
+- `agents[].class`
+- `agents[].model_hint`
+- `agents[].persistent_persona`
+- `agents[].reset_between_tasks`
+- `agents[].kickstart_template`
+
+**Verification:**
+
+Use `jq` for JSON examples or add a small Python/YAML validator if YAML is chosen. If no YAML parser is guaranteed, store examples as JSON or support both with Python stdlib JSON first.
+
+### Task 3.2: Add `mosaic fleet` commands
+
+**Objective:** Provide operator-safe commands for install/status/start/stop/restart/verify.
+
+**Files:**
+
+- Modify: `packages/mosaic/src/cli.ts` or the current commander entrypoint.
+- Create scripts under: `packages/mosaic/framework/tools/fleet/`
+
+**Commands:**
+
+```bash
+mosaic fleet init --profile minimal|usc --write
+mosaic fleet install-systemd
+mosaic fleet start [agent]
+mosaic fleet stop [agent]
+mosaic fleet restart [agent]
+mosaic fleet status --json
+mosaic fleet verify
+```
+
+**Implementation path:**
+
+Start by wrapping framework shell scripts from the TypeScript CLI. Do not overbuild a TypeScript service manager in the first pass.
+
+### Task 3.3: Add `mosaic agent` commands
+
+**Objective:** Provide transport-stable per-agent operations.
+
+**Files:**
+
+- Modify: Mosaic CLI entrypoint.
+- Create: `packages/mosaic/framework/tools/agent/` or reuse `tools/tmux` + `tools/fleet`.
+
+**Commands:**
+
+```bash
+mosaic agent roster [--json]
+mosaic agent status [agent] [--json]
+mosaic agent send <agent> --message "..."
+mosaic agent reset <agent> --clear|--new
+mosaic agent tail <agent> [-n 80]
+```
+
+**Reset behavior:**
+
+For tmux transport, `reset --clear` sends `/clear` then Enter through `send-message.sh`.
+
+For Claude/Pi differences, keep reset command configurable per runtime:
+
+```yaml
+runtimes:
+  claude:
+    reset_command: /clear
+  pi:
+    reset_command: /new
+```
+
+If a runtime does not support a known reset command, restart the service and send a fresh kickstart.
+
+---
+
+## Phase 4 — this-server rollout strategy
+
+### Task 4.1: Install on separate socket first
+
+**Objective:** Prove the holder pattern without disturbing existing sessions.
+
+**Commands after implementation lands locally:**
+
+```bash
+mosaic fleet init --profile minimal --write
+mosaic fleet install-systemd
+systemctl --user daemon-reload
+systemctl --user start mosaic-tmux-holder.service
+mosaic fleet verify
+```
+
+Expected:
+
+- `tmux -L mosaic-factory ls` shows `_holder`.
+- normal `tmux ls` still shows existing sessions unchanged.
+
+### Task 4.2: Start one canary agent
+
+**Objective:** Validate single-agent start/restart isolation.
+
+Use a harmless canary first, not the full fleet.
+
+Example roster addition:
+
+```yaml
+- name: canary-pi
+  runtime: pi
+  class: canary
+  working_directory: /home/jarvis/src
+```
+
+Commands:
+
+```bash
+systemctl --user start mosaic-agent@canary-pi.service
+SRV=$(tmux -L mosaic-factory display-message -p '#{pid}')
+systemctl --user restart mosaic-agent@canary-pi.service
+test "$SRV" = "$(tmux -L mosaic-factory display-message -p '#{pid}')"
+tmux -L mosaic-factory ls
+```
+
+Expected: holder PID unchanged; `_holder` remains; `canary-pi` recreated.
+
+### Task 4.3: Configure local Mosaic factory roster
+
+**Objective:** Create the actual local roster for this server after canary passes.
+
+Do not assume USC exact roster is desired here. Create a local profile such as:
+
+```text
+~/.config/mosaic/fleet/roster.yaml
+```
+
+Initial local recommendation:
+
+- `mos-claude` orchestrator
+- `coder0` / `coder1` implementers
+- `rev0` reviewer
+- `secrev0` security reviewer
+- `ultron` final/adversarial reviewer
+
+Scale to full USC-style pool only after resource/budget behavior is understood.
+
+### Task 4.4: Cut over existing ad-hoc tmux sessions only if desired
+
+**Objective:** Avoid data loss.
+
+Existing sessions on this server are not on the proposed `mosaic-factory` socket. They can remain untouched. If we later want them under Mosaic fleet control:
+
+1. list sessions;
+2. capture logs/handoffs;
+3. stop old processes intentionally;
+4. recreate as configured `mosaic-agent@...` services;
+5. verify comms and state.
+
+Do not run `tmux kill-server` on the default socket unless Jason explicitly approves that outage.
+
+---
+
+## Phase 5 — docs and AI Guide backfill
+
+### Task 5.1: Stack docs
+
+**Objective:** Document install and customization for Mosaic Stack users.
+
+**Files:**
+
+- Create: `docs/fleet/tmux-fleet.md` or `packages/mosaic/framework/tools/fleet/README.md`
+- Modify: top-level `README.md` if appropriate.
+
+Must cover:
+
+- what problem holder service solves;
+- install commands;
+- customization file;
+- example rosters;
+- reset/reuse lifecycle;
+- exact-target safety;
+- separate socket default;
+- Matrix migration path.
+
+### Task 5.2: AI Guide docs
+
+**Objective:** Keep generic guidance in AI Guide and implementation details in Stack.
+
+**Files in `mosaicstack/aiguide`:**
+
+- Update: `playbooks/tmux-fleet.md` with named socket, roster/profile, and resettable-slot pattern.
+- Add or update: `reference/agent-role-matrix.md` if PR #5 lands.
+
+Do not put Mosaic install commands as the only path in AI Guide. Present them as one implementation profile.
+
+---
+
+## Phase 6 — Matrix migration seam
+
+### Task 6.1: Add transport enum but implement tmux only
+
+**Objective:** Avoid hardcoding tmux into orchestration semantics.
+
+Roster:
+
+```yaml
+transport: tmux
+```
+
+Future:
+
+```yaml
+transport: matrix
+matrix:
+  homeserver: https://matrix.example
+  room_prefix: mosaic-factory
+```
+
+### Task 6.2: Define transport interface docs
+
+**Objective:** Make Matrix plugin work a transport swap, not a rewrite.
+
+Minimum operations:
+
+```text
+send(agent, message)
+reset(agent, mode)
+status(agent)
+tail(agent)
+listAgents()
+```
+
+Any tmux-specific concept must stay below this line.
+
+---
+
+## Acceptance criteria
+
+The implementation is complete when:
+
+- `mosaic fleet init` can write a minimal roster.
+- `mosaic fleet install-systemd` installs holder and agent units without hand editing.
+- `mosaic fleet start` starts the holder and configured agents on a named tmux socket.
+- Restarting one `mosaic-agent@name.service` does not change holder server PID or kill sibling sessions.
+- `mosaic agent send` can deliver a message to a named agent with a self-identifying preamble.
+- `mosaic agent reset` can clear/new a reusable slot and send a fresh kickstart.
+- `mosaic fleet verify` proves holder ownership, exact-target safety, and per-agent restart isolation.
+- Existing default tmux sessions on this server are not disturbed by default install.
+- Docs explain generic customization and include USC-style roster only as an example.
+- AI Guide remains generic; Mosaic Stack docs carry the concrete install path.
+
+## Risks and mitigations
+
+| Risk                                                | Mitigation                                                                        |
+| --------------------------------------------------- | --------------------------------------------------------------------------------- |
+| Killing existing tmux sessions                      | Use named `mosaic-factory` socket; no default `tmux kill-server`.                 |
+| systemd unit quoting/env expansion bugs             | Move logic into shell wrappers; verify with `systemd-analyze --user verify`.      |
+| Runtime reset command mismatch                      | Make reset command runtime-configurable; fallback to service restart + kickstart. |
+| Tool install drift                                  | Ensure npm package includes framework tmux/fleet tools; add packaging test.       |
+| Mosaic-specific assumptions leak into generic guide | Keep USC roster as example profile; AI Guide documents pattern/options.           |
+| Matrix migration blocked by tmux coupling           | Add `mosaic agent` abstraction now; keep tmux details below transport layer.      |
+
+## Suggested first PR split
+
+1. **PR A — tmux tool hardening**
+   - socket support;
+   - exact target helpers;
+   - tests/docs.
+
+2. **PR B — fleet systemd primitives**
+   - holder unit;
+   - agent unit;
+   - start-agent wrapper;
+   - install-user-units script;
+   - verify script.
+
+3. **PR C — roster and CLI**
+   - roster schema/examples;
+   - `mosaic fleet ...` commands;
+   - `mosaic agent ...` commands.
+
+4. **PR D — local rollout and docs**
+   - local roster for this server;
+   - run canary;
+   - document verification evidence;
+   - update AI Guide with generic lessons.
+
+## Immediate next action
+
+Implement PR A first. It is low-risk, improves existing tools, and is required for a safe named-socket rollout on this server.