Commit Graph
88 Commits
Author SHA1 Message Date
Dewey 9a5fbdbda7 fix(goal): quiet waits and unify fleet NG ownership (#56, #57, #58) 2026-09-06 04:07:09 -05:00
jason.woltje 7345f330fc docs: map foundation to integrated rewrite baseline 2026-09-06 02:40:05 -05:00
jason.woltje d4696d09eb feat(extensions): establish canonical goal source (#54, #55) 2026-09-06 02:32:32 -05:00
jason.woltje 44f257cb06 docs: record accepted phase-2 foundation contract 2026-09-06 02:23:15 -05:00
jason.woltje 69d1bb3aa4 docs(plan): resolve harness IDs + lifecycle review gate (#50)
Owner adjudication:
- canonical harness IDs match executables: pi, claude, codex, opencode
- agent.json uses one scalar harness ID; registry/manifest resolution, no
  hard-coded schema enum
- target mosaic harness list/detect/install/rm/status lifecycle
- detection recognizes reviewed executables and records compatibility
  without reading/copying harness homes
- installs are exact-version/verified, Mosaic-managed, never global
- detected external harness is available but not container-ready until
  imported/installed, absent a separately reviewed host adapter

Gate 1 resolved. Remaining P0/review gates stay open; no implementation
authorized. Suites 24/15/90/14/17 + verify green; unslop clean.
2026-09-04 12:26:57 -05:00
jason.woltje 9ea7dea711 docs(review): record independent glm-5.3 auth-registry spec review (#50)
Verdict ACCEPT WITH CHANGES. Persist ten-gate recommendations and three
P0 blockers: per-seat launch provider/model resolution; rotating OAuth
persistence for long-running seats; role auth ceiling ∩ settings profile
plus data-map/reset alignment. Reviewer made no repo edits.

Implementation remains blocked pending owner/conductor adjudication.
Prior gate remains green: suites 24/15/90/14/17 + verify.
2026-09-04 12:14:48 -05:00
jason.woltje d9a94f51ba docs(plan): specify harness declaration + centralized auth/provider registry (#49)
Design only; implementation blocked pending owner review.

- agent.json: one harness identifier (pi first), resolved through a
  versioned adapter/harness manifest; reusable settingsProfile reference
- central data-root registry: providers, accounts (metadata + secret
  credential split), reusable settings profiles, audited runtime selection
- per-seat pi auth.json/models.json mechanically generated and atomically
  activated; no seat/provider registration ceremony
- mixed oauth/api-key accounts supported centrally; one active account per
  provider per pi materialization
- host-side centralized OAuth login/refresh; agents never authenticate
- local/remote Ollama modeled as endpoint providers, not accounts
- target mosaic auth/provider/agent settings CLI; secrets never on argv
- migration, fail-closed acceptance suites, and ten explicit review gates

CURRENT.md points only to spec review. Suites 24/15/90/14/17 + verify
green; unslop clean.
2026-09-04 11:57:40 -05:00
jason.woltje 975084abe2 fix(auth): mosaic-managed auth lives under the data root, never ~/.pi (#48)
Owner direction: the stack must never impact default harness usage.
Correction to M19 as shipped (nothing had been created in ~/.pi — the
move breaks nothing).

- Mosaic-managed accounts: <dataRoot>/auth/<account>.json, perms 0600
  enforced (loose perms flagged in listings, refused by --auth — mirrors
  gitea-api.sh credential hygiene).
- ~/.pi is read-only to the stack, permanently; the only interaction
  remains the existing read-only container mount of the default
  credential. Recorded as a ROADMAP standing decision.
- auth.sh is now config-driven (data root from config.json, fail closed,
  consistent with every other tool); status reports both sources labeled.
- agent.sh --auth resolution moved after load_config (needs the data
  root); missing/symlinked/non-0600 accounts refuse.
- test-auth.sh: 15 no-Docker cases (accounts-create-nothing, loose-perms
  refusal, invalid-config refusal added). Test-authoring correction
  recorded in BUILD-LOG (fixture-state mismatch caught before running).

Suites 24/15/90/14/17 + verify green.
2026-09-03 22:53:33 -05:00
jason.woltje 073bbfdb6a feat(auth): M19 harness auth tooling — auth.sh checkpoint + per-launch account injection (#47)
Investigation (pi 0.84.4 docs + host auth.json metadata, values never
read): provider stacking is native (one auth.json keyed by provider;
resolution --api-key > auth.json > env > models.json; OAuth auto-refresh).
Multi-account per provider is NOT native -> named-file design:
auth.<account>.json + per-launch injection.

- scripts/auth.sh: status (provider names, credential types, perms,
  env-side names informational — never credential material) and accounts
  (named files, active marker). Exit codes per convention: 3 missing for
  a read, 2 unparseable, 4 file/environment (symlinks refuse).
- scripts/agent.sh --auth <account>: resolves auth.<account>.json and
  exports PI_AUTH_FILE (the existing compose read-only mount source — no
  new plumbing); missing/invalid account refuses pre-container.
- scripts/test-auth.sh: 13 no-Docker cases; core assertion is the safety
  property itself — fixture key/token/env VALUES never reach output.
- Docs: TOOLS.md Auth section, AGENTS.md command surface + suites.

Headless task runs keep the default credential (worker auth selection is
a separate policy decision). Real-host smoke: anthropic/openai-codex
oauth + zai api_key reported, perms 600, no named accounts yet.

Suites 24/90/14/17/13 + verify green. Agreed sequence M16-M19 complete;
M20 owner-gated.
2026-09-03 19:58:50 -05:00
jason.woltje d1d7b5598d feat(agent): fail-closed seat resolution under MOSAIC_AGENTS_DIR override (#46)
Owner decision after live verification of M18: an explicit agents-dir
override that cannot resolve the named seat now refuses the launch
(exit 4, names the seat and dir) instead of launching seatless and
unbounded. Unsetting the override keeps the M13 plain governed TUI.
MOSAIC_ROLES_DIR needs no symmetric change - the M18 gate already
refuses unresolvable role contracts.

Task suite 88 -> 90 (refusal + refusal-names-the-seat). TOOLS.md Agent
section documents the refusal.

Suites 24/90/14/17 + verify green.
2026-09-03 19:45:10 -05:00
jason.woltje ca8135d70c feat(roles): M18 seat-role progressive capability restriction (#45)
Role contracts (roles/<role>.json): roleVersion, name bound to filename,
tools ceiling (subset of pi built-ins), network declared (none|api-only|
open; enforced when network policy lands). Strict schema, fail closed -
a non-role document refuses resolution.

mosaic-task.mjs resolve-role: config-free contract validation, emits
MOSAIC_ROLE_TOOLS / MOSAIC_ROLE_NETWORK.

agent.sh: a declared role binds to its contract. Missing/invalid contract
refuses the launch (exit 2, names the role - the under-equipped-seat
failure mode, mirroring M17 skills). Effective tools = ceiling ∩ requested
(CLI --tools or agent.json caps); no request -> ceiling stands; narrowing
and tool-free outcomes loud on stderr. Adapters unchanged; headless M9
chain (mission ∩ task) untouched.

Ships roles/researcher.json (existing seat declares the role; without the
contract the fail-closed gate would refuse its launch).

Task suite 74 -> 88: contract resolution, wrong-kind/name/network/
duplicate/unsupported/missing refusals, ceiling narrowing E2E (mock
adapter), tool-free E2E, missing-contract refusal. Test-authoring
correction recorded in BUILD-LOG (a check that registered on one path
only, caught by count arithmetic).

Suites 24/88/14/17 + verify green.
2026-09-03 17:25:17 -05:00
jason.woltje bf56583a49 docs(skills): owner loop-doctrine + collaborator delivery discipline, remediated (#44)
ms-communications: integrated as-authored - owner preamble restructure +
collaborator delivery-discipline hunks from the #43 calibration (own
session output is not a send path; the tool performs the preamble flip;
receiving rule 3 requires actually running agent-send.sh).

ms-conductor: collaborator redraft integrated (canon-aligned tracking
surfaces, one-action cadence, fail-closed core) with one conductor
remediation - step 3 now distinguishes refusal (fail closed, never
bypass) from runner outage (direct dispatch to a qualified live seat via
ms-communications permitted, recorded loudly as degraded: no sandbox, no
run record; suites still gate integration). Preserves the owner's
outage-dispatch intent inside invariant 6.

docs/TOOLS.md: release.sh ensure row added (M16 subcommand existed in
code but not in the doc - flagged by the collaborator, verified in
release.sh usage).

Authorship: owner (ms-conductor doctrine, preamble restructure) +
ms-test collaborator (delivery hunks, redraft); remediation + integration
by conductor (dragon-lin:darkwing). Suites 24/74/14/17 + verify green;
unslop clean.
2026-09-03 17:17:29 -05:00
jason.woltje c8f433131c docs: calibration phase 22 record; CURRENT.md staleness corrected (M16/M17 late-logged, next M18); session registered (#43) 2026-09-03 16:55:31 -05:00
jason.woltje d1e75f855c docs(tools): document tools/ tree in TOOLS.md + fix stale suite counts (#43)
Collaborator-authored via conductor-loop calibration: task dispatched to the
live ms-test seat (glm-5.3-flash) over agent-send.sh; diff reviewed line by
line and every documented flag/exit code independently verified against tool
source by the conductor; suites green at integration (config 24 / task 74 /
release 14 / conductor 17 + verify).

- new 'Tools (host-side)' section: agent-send.sh, agent-watch.sh, unslop-check.js
- intro reading guide now points at tools/ (worker-flagged addition, accepted)
- Maintenance suite counts corrected: test-task.sh 58 -> 74

Authored-by: ms-test collaborator (glm-5.3-flash)
Integrated-by: conductor (dragon-lin:darkwing)
2026-09-03 16:55:25 -05:00
jason.woltje f710bf1a68 docs: KICKSTART file 2026-09-03 16:34:20 -05:00
jason.woltje 2085d75190 docs: ms-communications skill (owner-authored) + session registry entry + KICKSTART recovery file 2026-09-03 16:34:03 -05:00
jason.woltje 2f5a8d2cec feat(skills): skill lifecycle + ms-* skill set completion (#40, #41, #42)
- scripts/skill.sh: install (bundled or path) / activate / deactivate /
  uninstall (refuses while enabled) / list
- skills-enabled + skills-available dirs under the data root; a skill not
  in skills-enabled is not enabled or available for use
- pi adapter: MOSAIC_SKILLS -> --skill per dir; --no-skills when none
- agent.sh: seat definitions declare skills[]; resolution against
  skills-enabled refuses the launch loudly when missing
- ms-* skills completed (owner-authored canon, hands-off): ms-tools
  adapted to the runtime, ms-file-read/write/agent/conductor bodies
  written in the owner's style; ms-agent-watch + ms-unslop untouched
- tasks/USER.md onboarding fixtures; suite hardening (nested def path,
  user seed, mock-adapter dispatch evidence)

Suites: config 24, task 74, release 14, conductor 17, verify PASS.
RELEASE 0.0.12 packaged; health-gated activation on merge.

Closes #40, closes #41, closes #42
skill-lifecycle-v1
2026-09-03 16:23:57 -05:00
jason.woltje e14ad9ab52 Merge: skill lifecycle, skills completion, M16 self-determination hardening, M20 decision
- skill.sh lifecycle (install/activate/deactivate/uninstall/list)
- 8 ms-* skills completed (owner canon preserved)
- seat skills dispatch + enabled-dir resolution
- M16: ensure at launch, drift warnings, recursion guard
- M20 decision: packages/* monorepo at usurpation

Closes #40, closes #41, closes #42
2026-09-03 16:03:05 -05:00
jason.woltje 9fd16b9739 feat(release): recursion guard for the health gate; run-task drift warning; M20 packages/* decision recorded (#39)
- release.sh health gate runs with MOSAIC_ENSURE_SKIP=1: the gated task run
  cannot re-enter release self-determination
- run-task.sh warns on release drift instead of silently using a stale image
- ROADMAP: M20 decision recorded (packages/* monorepo at usurpation,
  continuity-first); restructure sequenced as M20 phase 1

Closes #39
2026-09-03 15:58:46 -05:00
jason.woltje 9051ad179b docs(roadmap): restructure sequencing (M20 phase 1, not first) + skills-as-discipline doctrine 2026-09-03 14:50:59 -05:00
jason.woltje 2f649ed930 docs: README release ensure row 2026-09-03 14:43:25 -05:00
jason.woltje db330c12c7 feat(release): self-determination - ensure at launch, drift warnings, M20 packages/* decision (#38)
- release.sh ensure: aligned no-op; drift -> package-if-needed + health-gated
  activate (M3 gate-then-flip, automated)
- ensure_release_aligned in common.sh: invoked by hello/verify/agent;
  MOSAIC_ENSURE_SKIP guards recursion; run-task warns on drift without
  auto-aligning (workers/suites never trigger builds or model gates)
- ROADMAP: M20 decision recorded - v2 adopts packages/* monorepo at usurpation
- BUILD-LOG Phase 20 + tool-race process note

Live-verified: post-reset pointer loss auto-restored via health-gated
ensure; drift warning fires on desired-version bump; idempotent no-op on
aligned state.

Closes #38
2026-09-03 14:42:09 -05:00
jason.woltje 0a17d29bce docs(roadmap): flush 2026-09-03 14:28:58 -05:00
jason.woltje 3458f6a7ad docs(roadmap): M17 skill lifecycle (skills-enabled/available, role-scoped subsets, --skill negates --no-skills); M20 stack succession path + monorepo question 2026-09-03 14:28:45 -05:00
jason.woltje b033952cd5 docs(harvest): pattern ledger from stack/next + fleet runtime (12 patterns, skips, owner-corrected M17-M19 designs) 2026-09-03 13:55:30 -05:00
jason.woltje c102980ad4 docs(plan): CURRENT.md - queue aligned to ROADMAP (M16 next, CI deferred) 2026-09-03 12:59:28 -05:00
jason.woltje 58b96cb715 docs(plan): ROADMAP.md - M16 release self-determination, M17 ms-tools skill, M18 seat-role restriction, M19 auth tooling; CI deferred per owner 2026-09-03 12:57:26 -05:00
jason.woltje dd3ff944a1 chore(release): 0.0.11 onboarding-v1 2026-09-03 12:30:52 -05:00
jason.woltje 8eb81ebec1 feat(onboard): user onboarding - no default USER.md, guided creation (#37)
- bootstrap no longer creates user/USER.md (owner direction)
- scripts/onboard.sh: name REQUIRED (interactive loop or --name),
  optional fields prompted (profession, marital, age, gender, education,
  location, timezone, skillset, interests, hobbies, pets); flag-driven
  non-interactive mode for automation
- templates/USER.md: canon skeleton, placeholder rendering, unfilled
  optional = (not provided)
- agent.sh: auto-runs onboarding when profile missing (TTY gate);
  headless run-task warns and continues without user context
- user profile dispatched to all launches (M14 layer)

Closes #37 (onboarding requirements from owner layout review)
2026-09-03 12:30:42 -05:00
jason.woltje 530597cc84 docs(plan): CURRENT.md flush 2026-09-03 11:59:49 -05:00
jason.woltje 9a0d44f96a docs(plan): CURRENT.md - deduplicated completed log (marked correction), M15 review queued 2026-09-03 11:59:37 -05:00
jason.woltje 121b331c6c docs(log): back-fill phases 16-19 (M10, M12, M14, M15) - recorded retroactively with ground-truth sources 2026-09-03 11:59:14 -05:00
jason.woltje 34e06e7de7 Merge M15: agent seats - per-agent SOUL and role contracts
Closes #36
agent-seats-v1
2026-09-03 11:56:38 -05:00
jason.woltje 9bd4f1c405 feat(agents): agent seats - per-agent SOUL, role, definitions dir (#36)
- agents/<name>/ holds agent.json (strictly validated: version, name,
  role?, capabilities?, workspace?, session?) + SOUL.md (persona prose)
- agent.sh: definition loading (quote-safe node defaults file), runtime
  SOUL copy to dataRoot/agents/<name>/, MOSAIC_AGENT_SOUL_FILE ->
  loader fills the SOUL slot from the seat's persona (contract SOUL =
  default persona; governance never overridden)
- seat.json written once at instantiation (seatVersion, name, role, at)
- identity section gains agent role; compose passthrough for role+SOUL
- live user context (M14) + seat SOUL compose the full persona:
  governance -> persona -> identity -> user -> mission
- RELEASE -> 0.0.10; packaged and health-gated activated
- example seat committed: agents/researcher

Closes #36
2026-09-03 11:56:38 -05:00
jason.woltje a7b612435b docs: BUILD-LOG Phase 15, CURRENT.md - M13 shipped 2026-09-03 11:25:41 -05:00
jason.woltje 87f10772ce Merge M13: interactive TUI agent + TOOLS.md
Closes #35
interactive-agent-v1
2026-09-03 11:24:56 -05:00
jason.woltje 7db4c5c2ed feat(agent): interactive TUI launcher + identity + TOOLS.md (#35)
- scripts/agent.sh <name>: launches interactive pi TUI in the container
  with contracts + optional mission + agent identity + named session +
  optional workspace/tools; the Mosaic alternative to vanilla pi
- pi adapter: MOSAIC_INTERACTIVE branch (clean TUI, no -p, no initial
  prompt); headless exec rebuilt via positional args (no word-splitting
  on the request); MOSAIC_AGENT_NAME optional in headless
- loader: AGENT IDENTITY section when the launcher names the agent
- compose: fixed command removed (request defaults live in run-agent.sh);
  MOSAIC_INTERACTIVE/MOSAIC_AGENT_NAME passthrough
- docs/TOOLS.md: full on-demand tool reference; AGENTS.md routes to it
- RELEASE -> 0.0.8 (container change); build verified

Closes #35
2026-09-03 11:24:56 -05:00
jason.woltje 0273a84549 docs: AGENTS.md - session recovery shim, invariants canon, session registry
- AGENTS.md at root: pi loads it automatically at every session start
  (conductor-level sessions; workers deliberately exclude it via
  --no-context-files). Deliberately short: invariants, session protocol,
  role model, command surface, data map, pointers - depth stays in docs/.
- docs/SESSIONS.md: append-only session registry, mandatory per session.
- Recovery rule encoded: compaction/restart loses nothing - AGENTS.md +
  CURRENT.md + git log + suites reconstruct state; never guess.
2026-09-03 11:02:19 -05:00
jason.woltje 3b674b7a66 Merge: roles/ directory convention - root is bootstrap-only 2026-09-03 10:56:07 -05:00
jason.woltje 527bc581ca refactor(layout): role contracts move to roles/ - root is bootstrap-only
Owner direction: the repository root holds first-class, bootstrap-required
configuration only. conductor-policy.json is a ROLE contract (the
conductor's authority), one of scores of future role contracts
(agent-policy, coder-policy, ...) - such files get a dedicated home.

- roles/conductor-policy.json (git mv)
- conductor-apply.sh + test-conductor.sh read the new path
- CONDUCTOR.md records the roles/ convention

Closes UX follow-up from owner layout review; no issue (convention change).
2026-09-03 10:56:07 -05:00
jason.woltje 1249714a9a docs(plan): CURRENT.md - M12 shipped 2026-09-03 07:04:37 -05:00
jason.woltje e175616885 feat(conductor): auto-apply policy gate for worker patches (#34)
- conductor-policy.json (tracked, strictly validated): enabled switch,
  path allowlist globs, gating suites - the autonomy decision lives in a
  declarative file the owner controls
- scripts/conductor-apply.sh <runId> [--dry-run]: succeeded-run check ->
  clean target tree -> diff from worker workspace -> allowlist -> syntax
  gates (node/bash/json) -> apply -> policy suites -> attribution commit;
  ANY failure reverts the tree; push is never automatic
- scripts/test-conductor.sh: 17 sandbox cases covering every gate incl.
  suite-failure auto-revert and disabled policy
- policy defaults: scripts/docs/tasks/missions/adapters + README; all
  three suites gate

Closes #34
auto-apply-v1
2026-09-03 07:03:29 -05:00
jason.woltje 6955717612 Merge M11: session forking from a common ancestor
Closes #33
session-fork-v1
2026-09-03 06:43:02 -05:00
jason.woltje 88d9cf750f feat(sessions): sessionForkFrom - branch conversations from a common ancestor (#33)
- task schema: optional sessionForkFrom (source session name); requires
  session target; self-fork rejected
- runner: resolves source newest .jsonl (fail 4 if none/outside dataRoot);
  passes MOSAIC_SESSION_FORK + MOSAIC_SESSION_DIR; result records lineage
- pi adapter: --fork <source> --session-dir <target> when forking;
  ephemeral default unchanged; plain session resume unchanged
- compose passthrough; RELEASE -> 0.0.7 (adapter changed)
- suite +9 cases (58 total): plumbing via mock stderr, validation
  negatives, live fork - child recalls ancestor code word, ancestor
  session file untouched

Closes #33
2026-09-03 06:43:02 -05:00
jason.woltje c038706eed docs(plan): CURRENT.md - M10 shipped, retention next in review 2026-09-03 06:33:54 -05:00
jason.woltje 8622c9d826 Merge M10: run-record retention
Closes #32
retention-v1
2026-09-03 06:33:27 -05:00
jason.woltje 88eef507b0 feat(retention): run-record pruning - keep newest N, dry-run default (#32)
- mosaic-task.mjs prune [--keep=N] [--yes]: default keep 50; without
  --yes lists candidates without deleting
- only r-* directories under the runs root; symlinks skipped;
  sessions/workspaces/state/config untouched (asserted by suite sentinels)
- append-only receipt runs/.pruned.log records every pruned id
- test-task.sh: +8 retention cases (dry-run no-delete, keep-N, newest
  kept, receipt, isolation, invalid keep, empty no-op)

Also: suite hardening - prune section scopes its config per-command
(no export/unset leaking into later sections); duplicated check()
removed; latest_reason hoisted to helpers; status colors now green OK /
red FAIL (terminal-only, NO_COLOR-aware) per owner UX feedback.

Closes #32
2026-09-03 06:33:27 -05:00
jason.woltje 439bea6915 ui(test): green OK/PASS, red FAIL - terminal-only, NO_COLOR-aware
Owner feedback: grep match-highlighting made the word 'policy' red while
status words were plain - counter-indicative. Suites + verify now emit
ANSI colors (green success, red failure) when stdout is a terminal;
piped/machine-parsed output stays plain, honoring NO_COLOR. Word 'ok'
promoted to 'OK' for scannability.

Verified byte-level via forced-pty run; piped output unchanged; suites
41/24/14 + verify green.
2026-09-03 06:23:47 -05:00
jason.woltje fad8a4718c Merge M9: mission-level capability policy
Closes #30
mission-policy-v1
2026-09-03 06:16:01 -05:00
jason.woltje 2ff49adff4 feat(policy): mission-level capability policy - least-privilege intersection (#30)
- mission schema: optional capabilities.tools (same validation as task)
- merge semantics in runTask: neither -> none; mission only -> mission;
  task only -> task; both -> intersection (task narrows, never widens);
  empty intersection -> tool-free run with an explicit stderr note
- result.json records EFFECTIVE tools; task/mission snapshots remain the
  immutable declaration of intent
- adapters unchanged; host-side only (no image change, 0.0.6 still active)
- task suite +5 cases (41 total): all four merge cases asserted from run
  evidence + invalid mission capabilities rejected

Policy decision recorded: missions govern; tasks cannot escalate.

Closes #30
2026-09-03 06:16:01 -05:00