mirror of
https://github.com/rennf93/roboco.git
synced 2026-08-03 07:23:24 +02:00
236aab18f564d1d4b624146c2fb1de9799c2d79b
7
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
3849c1737e |
feat(video): switch 0.19.0 renderer Remotion → HyperFrames (HTML-native, Apache-2.0) (#314)
* feat(video): rewrite sidecar render core to HyperFrames (in place)
* feat(video): convert motion compositions from Remotion TSX to HyperFrames HTML
* refactor(video): rename render client to video_renderer_client (renderer-agnostic)
* chore(video): rename remotion-renderer prose in test_video_pipeline docstrings
* chore(video): rename sidecar to video-renderer + add system ffmpeg for HyperFrames
* chore(video): rename stray remotion-renderer refs in sidecar + py docstrings (controller cleanup)
* chore(video): fix stale Remotion API names in Dockerfile comment (controller cleanup)
* docs(video): rewrite video-engine prose for HyperFrames + add map entry + folded prose fixes
* docs(video): add trailing newline to docs/map/video-engine.md (controller cleanup)
* chore(video): drop internal spec refs + minio/test suppressions (folded hygiene)
* fix(video): reclaim outDir on createRenderJob throw + hide empty 4th highlight
Final whole-branch review (Opus) triaged two FIX items from the SDD nits
ledger; the rest ship as-is.
- render.js: a synchronous throw from createRenderJob (post-mkdtemp, not
awaited) left an empty outDir on disk — the outer catch only reclaimed
extractDir. Reclaim outDir too when it exists, and correct the stale
comment that claimed the out dir was never created.
- {vertical,square}.html: the 4th highlights <li> lived in the DOM hidden
only by JS, so a no-JS / failed-script render would show an empty bullet.
Start it style="display:none" and reveal on populate, so an unscripted
render shows nothing instead.
Vitest smoke (release-announcement.test.js) 4/4 green; render.js syntax
checked. Python suite untouched by this fix (JS/HTML only).
* fix(video): type _override_db yield as AsyncSession | None
T7 widened _build_app's db_session param to AsyncSession | None (to drop the
4x # type: ignore[arg-type] on the DB-independent _build_app(None, ...) calls)
but left the inner _override_db fixture typed AsyncIterator[AsyncSession] —
so 'yield db_session' yielded AsyncSession | None into a declared AsyncSession,
and mypy failed at test_video_routes.py:177 ('Incompatible types in yield').
The DB-independent media tests pass db_session=None deliberately: their route
uses a monkeypatched task service and never awaits the session, so yielding
None is safe at runtime. Type the override's yield as AsyncSession | None to
match — no cast, no # type: ignore, no assert, runtime behavior unchanged.
The 3 media tests (3 passed) and the 19 db-gated tests (skipped locally) hold.
* chore(gate): skip .superpowers scratch in markdown prose gate
reflow_md.py walks the filesystem via rglob('*.md') and skips tooling dirs
(.venv, .mypy_cache, .pytest_cache, ...) but not .superpowers/ — the
superpowers SDD workflow's scratch dir (briefs, reports, progress ledger,
all gitignored). A dev running SDD locally would hit a false markdown-prose
gate failure on those transient files. Add .superpowers to SKIP_DIRS,
consistent with the existing tooling-scratch exclusions.
* fix(video): validate composition_id to close path traversal (CodeQL)
compositionId flowed unvalidated from the POST body into path.join
under extractDir/motion/compositions/, so a '../..'-style value could
escape the composition dir (CodeQL: Uncontrolled data used in path
expression). Validate at the trust boundary in server.js
(/^[A-Za-z0-9_-]+$/) and add a path.resolve + startsWith containment
check in render.js so it stays safe regardless of caller.
* fix(mcp): send X-Agent-Token + X-Agent-Team from flow/do servers
flow_server._build_headers and do_server._build_headers constructed
only X-Agent-ID/Role/Correlation-ID, omitting X-Agent-Token and
X-Agent-Team (unlike ApiClient._get_agent_headers used by the other
MCP servers). Latent since the gateway refactor — surfaced when
ROBOCO_AGENT_AUTH_REQUIRED=true was armed on the NAS, 401-ing every
flow/do verb with 'Missing X-Agent-Token header'. Add both headers
(mirroring ApiClient) so the HMAC gate passes. Tests assert the
headers are now injected.
* [video-engine] Per-project video_engine_enabled opt-in toggle
Mirrors ci_watch_enabled (migration 048): the global
ROBOCO_VIDEO_ENGINE_ENABLED flag arms the subsystem; the new
projects.video_engine_enabled column (migration 063) opts a repo into
authoring against its motion/ dir. VideoEngine._opted_in_project no-ops
open_video_task at the single chokepoint covering all three trigger
paths (on-release, on-spotlight, CEO on-demand) until the operator
flips it in the panel edit-project dialog. Existing projects stay
opted out (server_default=false).
* fix(auth): send X-Agent-Token + X-Agent-Team from all agent->API call sites
The prior fix (
|
||
|
|
0ca9d91b72 |
v0.16.0: fastapi-guard HTTP security layer — calibrated + scanner honeytrap (#290)
* [fastapi-guard] Phase 1a: gated config flags for the HTTP security layer Adds the ROBOCO_GUARD_* settings (all default-off / secure-default) for the upcoming fastapi-guard 7.2.0 hardening — guard_enabled (master switch), guard_fail_secure (fail-closed default; NAS overrides to false), guard_telemetry_enabled + guard_agent_api_key + guard_project_id (guard-agent telemetry, opt-in), guard_emergency + guard_emergency_whitelist (lockdown kill switch). Inert until consumed: nothing reads them yet, so the request path is unchanged. Foundation for v0.16.0. * [fastapi-guard] Phase 1b: security foundation module + gated wiring Add fastapi-guard 7.2.0 + guard-core 3.3.0 (bare, unpinned) and roboco/security.py: - build_security_config() from settings — behind-nginx real-IP (trusted_proxies + trust_x_forwarded_proto), HSTS/CSP headers, threat-ban + 404-sweep rules, redis-backed state, exclude_paths (/ws + health + docs), env-driven enforce_https, fail_secure (secure default), emergency lockdown, guard-agent telemetry (opt-in), passive-mode calibration switch. - guard_deco singleton (SecurityDecorator) for per-route decorators (Phase 2+). - Three custom content validators guard's WAF can't cover: prompt-injection / role-override, secret-exfil / credential-in-body, internal-SSRF. - apply_guard(app) + guarded_lifespan() wired into create_app AFTER settings. guard_passive_mode config flag added. Entirely gated by ROBOCO_GUARD_ENABLED (default off): create_app mounts nothing and returns the unchanged app when off (verified). make quality GREEN (cov 95.32%, pip-audit clean, import-linter 2/0). 12 new unit tests. * [fastapi-guard] Phase 2: critical-path decorators Apply guard decorators to the highest-value endpoints (metadata-only; enforced only when the middleware is mounted, so no-op while ROBOCO_GUARD_ENABLED is off): - provider keys (ollama/grok/self-hosted writes): strict rate_limit + max_request_size + block_clouds (no datacenter IP should touch secret writes). - settings write + release approve/reject (CEO-gated): strict rate_limit. - intake chat (prompter start/messages/events): rate_limit + max_request_size + custom_validation(prompt_injection_validator) — the prompt-facing free-text ingress gets the injection/role-override/secret-exfil content scan. make quality GREEN (cov 95.32%, contracts 2/0). App builds with guard off, decorators inert (verified). * [fastapi-guard] Phase 3: wide decorator coverage across ingress + sensitive routes Targeted-wide application (metadata-only; no-op until ROBOCO_GUARD_ENABLED). The global SecurityMiddleware already rate-limits + WAF-scans every request, so this adds the custom content validators on free-text ingress + tight limits on sensitive ops (not blanket per-route rate_limit on reads): - agent gateway do verbs (note/say/commit/dm/pitch/progress/draft_playbook/...): rate_limit + max_request_size + custom_validation(secret_exfil or prompt_injection). - a2a message/send + chat writes: rate_limit + size + prompt_injection. - optimal/RAG (kb/search, rag/query, mentor/ask, errors/decisions/standards/ learnings): prompt_injection on searches, secret_exfil on record writes; docs index → internal_ssrf. - tasks: create/update → prompt_injection; QA/doc/PM transitions → secret_exfil; CEO-gated verbs → tight rate_limit. - secretary chat → prompt_injection; research → internal_ssrf; orchestrator spawn/mutations → rate_limit; git ops + flow verbs → tight rate_limit. Pure GET/reads left to the global middleware. Applied via a Sonnet workflow, then verified: app builds with guard off (decorators inert), make quality GREEN (cov 95.37%, contracts 2/0). Decoy/honeypot-path surface deferred (needs verified guard ban-API integration — not rushed). * [fastapi-guard] Phase 5: arm the NAS composes in passive/log-only mode Arm ROBOCO_GUARD_ENABLED=true + ROBOCO_GUARD_PASSIVE_MODE=true + ROBOCO_GUARD_FAIL_SECURE=false on the two NAS composes (docker-compose.yaml + .yml). Passive = guard mounts and logs what it WOULD block but blocks nothing, so the next NAS deploy calibrates against real traffic; flip PASSIVE_MODE off after the false-positive review to enforce. fail_secure=false keeps a guard-internal error from 500ing the personal deploy. The registry (user-facing) compose is deliberately left unarmed so its published default stays conservative. Phase 4 (passive calibration) is the operational step this enables. * feat(security): Phase 3b — full-arsenal per-route guard enrichment Stack the applicable guard decorators per surface instead of the minimal rate_limit/max_request_size/custom_validation triad: content_type_filter on every JSON-body write, honeypot_detection form-traps on human-facing POSTs, block_clouds on key-writes + CEO release ops, behavior_analysis runaway-rate rules on the agent flow/do verbs, suspicious_detection + usage_monitor on the sensitive surfaces. Nine distinct decorators now applied thoughtfully per endpoint. All metadata-only — no-op while ROBOCO_GUARD_ENABLED is off. * fix(a2a): permit PR reviewer to deliver gate verdicts to the owning PM can_a2a_direct had no pr_reviewer rule, so a reviewer (team=board, or a cell team) fell through to the cell-member path and was cross-cell-denied when the in-path gate delivered a pr_fail change-request to main-pm (or a cross-cell cell-pm): "Cannot A2A main-pm ... Ask None to coordinate with None". The delivery is best-effort, so pr_fail still transitioned but the verdict never reached the owning PM — the blind-re-submit signal-gap the pr_fail fix closes. Add an explicit pr_reviewer handler: it may A2A only cell_pm / main_pm (its sole comms surface — everything else it posts on the PR itself), with a matching route hint. The cell reviewers kept same-team access by coincidence; this scopes every reviewer to PM-only, the correct model, with no other A2A caller affected. Refresh uv.lock to the current resolution. * feat(models): adopt Claude Sonnet 5 as the sonnet tier Point the 'sonnet' alias at claude-sonnet-5 (MODEL_MAP) and give pr_reviewer its own opus tier in ROLE_MODEL_MAP — it was falling through to the sonnet default, and the role gates untrusted external/fork PRs plus root→master, which warrants opus. Price claude-sonnet-5 at the promotional 33% off Sonnet 4.6 ($2.01 / $10.05, cache 0.201 / 0.5025) through 2026-08-31 via a dedicated pricing fragment that beats the bare 'sonnet' alias; revert to full rate when the promo ends. Bare 'sonnet' stays full-rate as a conservative fallback (prod prices the resolved claude-sonnet-5 id from the transcript). Update the model docs and the billing / usage / manifest / spawn tests. * feat(security): calibrate the guard WAF for RoboCo traffic + document the layer The first end-to-end run of the fastapi-guard layer showed active enforcement would block ~50% of legitimate agent traffic — RoboCo request bodies are code, SQL, diffs, file paths, HTML, and URLs, which the stock signature WAF reads as attacks. build_security_config now excludes RoboCo's free-text top-level body fields (derived from the real request models, including the free-form container fields whose nested prose is stringified and scanned) from WAF scanning, dropping the active-mode false-positive rate to zero while keeping the WAF on every non-excluded (id/enum/slug/branch) field and leaving the prompt-injection / secret-exfil / internal-SSRF validators — which run independently of the exclusion — fully in force. enable_penetration_detection is made explicit. Only excluded_detection_body_fields is reliable on guard 7.2.1: the per-route categories knob is bypassed for JSON bodies, and the body scanner excludes top-level keys only (scanning str(value) of every non-excluded field), so free-form container fields must be excluded wholesale. Adds tests/unit/test_security_middleware.py — the first end-to-end exercise of the middleware (mounts it, drives guard's lifespan, fires real requests): proves passive mode is log-only, active mode does not false-positive on realistic agent payloads, threats are still blocked inside excluded fields, and the WAF still fires on non-excluded fields. Docs: CHANGELOG (Unreleased); a user-facing Optional-subsystems page + nav + env reference for the HTTP security layer; the agent-facing RAG corpus (what it is + why a request could be blocked); and the roboco mapping (api-core-websocket / deployment-tooling / _complete_map). * feat(security): Surface N — scanner honeytrap (guard /api auto-ban + nginx edge-drop) Turns scanner probes against the scanner, in two layers matched to where traffic lands. Behind nginx only /api, /ws, /health, /ready reach the orchestrator, so guard can only see (and ban) scanner probes on those paths; the classic root probes (/.env, /wp-login.php, /phpmyadmin, /.git/config) hit the panel. So: - build_security_config's threat_ban_config gains recon / sensitive_file / cms_probing categories. A scanner probing those fingerprints on an /api path is detected on the URL-path scan; repeated probes from one IP trip an adaptive per-IP auto-ban (redis-backed, 24h). Only bans in active mode (passive logs the recon hit) and needs redis (the 24h ban exceeds the in-memory cap). The spec's decoy-route file is redundant — the WAF url-path scan bans regardless of a registered route — so it is intentionally omitted. - docker/nginx.conf drops the classic root scanner paths at the edge with 444 (connection closed, no response) before they reach the panel, anchored to known scanner fingerprints so /.well-known and every real panel/API route are untouched. Always on, independent of ROBOCO_GUARD_ENABLED. Tests: 2 unit (the exclusion set + the scanner-ban categories are present) and 2 integration (a decoy path is blocked in active mode, passes in passive). The nginx regex was validated against 15 scanner + 19 legit paths (0 false positives). Docs: CHANGELOG, the HTTP-security page, the roboco mapping, and the agent-facing RAG corpus. * Token optimization — per-role observability, compute policy, spawn preflight (#291) * test(models): lock the sonnet→claude-sonnet-5 MODEL_MAP invariant * feat(usage): surface cache tokens + cache_hit_rate in usage breakdowns * feat(usage): add per-role usage breakdown endpoint * feat(usage): add spawn-waste signal (per-role unproductive rate + respawn strikes) * feat(panel): surface per-role cost/cache + spawn-waste on the metrics page * feat(routing): Phase 2 per-role compute policy — qa→haiku, main_pm→sonnet, per-role effort env mechanism (default-inert) * feat(orchestrator): Phase 3 flag-gated spawn preflight — refuse non-gateway delivery roles (respawn-forever guard) * chore(compose): arm ROBOCO_SPAWN_PREFLIGHT_ENABLED on the NAS composes * docs: per-role usage observability, per-role compute policy, and spawn preflight --------- Co-authored-by: Renn F <rennf93@users.noreply.github.com> * fix(panel): pin outputFileTracingRoot so the standalone build isn't broken by stray lockfiles * feat(routing): populate ROLE_EFFORT_MAP + wire the verified --effort flag (cell_pm/board/auditor to medium) * feat(gateway): omit empty context_briefing sections (Phase 4 payload compaction) * refactor(orchestrator): extract spawn chokepoint guards to restore xenon rank B on spawn_agent --------- Co-authored-by: Renn F <rennf93@users.noreply.github.com> |
||
|
|
536bbb64f3 |
Chore/all/logical gaps sweep (#286)
* release-manager: fencing-token mutex + executor/readiness hardening
Closes the release-mutex TTL race (#17, HIGH) and the remaining
release-manager gaps (#88, #89, #201, #202):
- #17: the release mutex is now acquired with a uuid4 fencing token and
released via Lua compare-and-del; a background asyncio heartbeat
compare-and-expires the TTL ~every 60s while the execute owns the lock,
so a live execute no longer expires and a crashed one auto-releases
<=3000s. A second approve after TTL expiry cannot usurp and rm -rf the
in-flight clone — the fenced first-finally keeps its lock.
- #89: a Redis outage during acquire stays fail-closed (the execute never
runs) but now returns a distinct redis_unavailable result + log so the
CEO sees the cause instead of a false already_in_progress.
- #88: commit_and_push RuntimeError is wrapped into a structured
ReleaseResult(commit_failed) instead of a 500.
- #201: first-release fallback still emits untracked version-ref files as
gaps (no longer silenced by the first-release branch).
- #202: _await_proc awaits proc.wait() after kill() so a timeout cannot
leak a zombie.
TDD: tests/unit/services/test_release_proposal_concurrency.py extends
_FakeRedis with eval/get/expire and pins the fencing/heartbeat/usurper
invariants + the redis_unavailable result.
* PM/code-task creation guard + main_pm coverage + issue carve-out
Closes the creation-time role x task_type gap (the user's explicit example)
and the main_pm delegate hole:
- New pure helper `pm_cannot_own_code(role, task_type, is_issue_resolution)`
in foundation/policy/batch.py — single source of truth. Both PM roles
(cell_pm + main_pm) coordinate; a `code` task assigned/claimed by a PM is
a structural mismatch, EXCEPT a PM taking a code task in needs_revision to
resolve review/QA issues directly (the carve-out).
- Creation-time guard: TaskService.create calls the helper (closes the
create-with-cell-PM-assignee hole the team-based check misses).
- Delegate path + spec claim gate consult the same helper.
`_validate_assignee_task_type` / `_task_type_hint_for` now key on the
Role (CELL_PM OR MAIN_PM), not the cell-PM slug set — closes the
delegate-to-main-pm-as-code hole.
- identity.role_for_uuid_or_none is None-tolerant (treats None as "not a
PM" and proceeds) so a malformed/missing assignee cannot crash the guard.
- prompter.create_task_from_draft reuses the guard at draft-create.
TDD: test_batch.py (helper matrix + carve-out), test_main_pm_code_guard.py
(main_pm coverage), test_delegate_assignee_task_type.py (delegate parity),
test_lifecycle_spec.py (claim gate: rejects PM claiming code from pending,
allows from needs_revision + PM claiming planning + dev claiming code).
* task-service: completion hooks + escalation/cancel/audit hardening
Closes the task-service cluster (#21/#98, #99, #100, #101, #103, #216; #102
verified already-covered, #217 verified already-guarded):
- #21/#98: ceo_approve now closes the work session + triggers completion
hooks before worktree removal (no-op when work_session_id is None), so a
CEO-approved task lands the same close-path as PM-completed.
- #99: apply_escalation routes through the transition validator with an
enumerated escalation exemption (_ESCALATABLE_TO_BLOCKED) instead of an
arbitrary source->BLOCKED write; BACKLOG is refused.
- #100: branchless ceo_reject awaiting_ceo_approval->pending gets a real
spec edge (ceo_reject_to_pool ActionSpec + _STATUS_TRANSITIONS entry) so
future admin-override tightening can't wedge the path.
- #101: revision_count bump is documented as the single chokepoint, with
the pre-block RESTORE path undoing it when restoring a snapshotted
needs_revision (same cycle resuming, not a new rejection).
- #103: cancel cascade surfaces non-terminal orphans instead of swallowing
the role violation.
- #216: _remove_task_worktree_on_terminal escalates recurring FS/permission
failure (audit/notify after N) instead of silent-failing forever.
- #102: pinned in test_verb_runner_midverb_invalid_state.py (committed with
the choreographer cluster) — verb-runner savepoints already surface a
concurrent mid-verb state change as INVALID_STATE.
- #217: submit_for_qa claimed_by guard verified intact.
TDD: test_task.py, test_worktree_cleanup_on_complete.py,
test_escalation_board_guard.py (#99), test_task_service_* integration,
test_lifecycle_spec.py.
* choreographer: gate-claim guards + pr-gate hardening + fail-open logging
Closes the choreographer cluster (#5/#222, #29, #30, #82, #188, #189,
#192; #157/#187 verified already-fixed/pinned; #102 pin lives here):
- #5/#222: the unchanged-PR guard's fail-open head_sha lookup now logs
(warning) on a slug-resolver/git-helper error so a regression cannot
silently turn the pr_fail re-submit loop-stopper into a no-op. Stays
fail-open (never wedges the PM).
- #29: pinned (REFUTED-with-pin) — _lane_claim_guard already returns the
error envelope without releasing the claim on a transient lookup error.
- #30: pinned (REFUTED-with-negative-pin) — a non-batch branchless main_pm
root cannot bypass the complete spec gate (is_batch_umbrella requires
batch_id set).
- #82: _post_gate_review_to_pr wraps the slug-resolution call in try/except
(mirrors _capture_pr_head_sha) so a malformed cell_map AttributeError no
longer 500s the reviewer after a committed gate transition.
- #188: _is_hand_formatted_verdict anchors the header regex to line-start,
so a quoted (> ## Summary) or inline (mid-prose) header mention no longer
false-refuses a hand-formatted verdict.
- #189: pr_fail re-captures the PR head SHA after the transition commits and
re-stamps the verdict note only when it advanced (closes the stale-SHA
false-allow loop-hole); no-advance stays a single note write.
- #192: claim_gate_review skips the dev claim guards (already_active/paused/
lane) via a new skip_dev_guards param — a pr_reviewer inspecting an
assembled PR does not start work, so the single-active-task / code-lane
invariants do not apply; the dependency guard is kept, and QA's
claim_review parity is preserved.
- #157/#187: verified in tree — pr_review-only handoff is intentionally
prior-work-worth-resuming; self_review_block wiring (reviewer != dev)
holds on assembled tasks with 4 existing pin tests.
TDD: test_choreographer_*, test_pr_gate_posts_review (#82),
test_pr_review_hand_format_guard (#188), test_submit_root_unchanged_pr_guard
(#189), test_claim_gate_review_guards (#192), test_verb_runner_midverb
_invalid_state (#102 pin).
* playbook curate: guard the gating commit against a poisoned session (#55)
The explicit `session.commit()` that gates the RAG index (commit-before-index
so an uncommitted playbook cannot land in the corpus) raised PendingRollbackError
when a prior mid-verb failure had rolled the caller's session back — 500ing the
whole curation verb instead of returning a clean envelope, and (worse) risking a
fall-through to index an uncommitted playbook. Wrap the commit: on
PendingRollbackError, log + return invalid_state with a re-fetch/retry remediate
and skip the index. The happy path still commits exactly once then indexes.
TDD: test_playbook_verbs.py — poisoned-session returns a clean invalid_state and
does NOT index; clean-session still commits once + indexes (pins no fail-closed
inversion / no double-commit).
* [chore] gateway: atomic activate merge — preserve probe_failures across re-park (#156)
activate() was a blind SET that reset probe_failures to 0, so a probe-failure
increment that just landed (or was in flight) could be wiped by a concurrent
re-park — resetting the give-up / CEO-notify count mid-episode. Route activate
through a server-side Lua merge (roboco:activate_rate_limit) that refreshes the
episode metadata (kind / activated_at / retry_after / affected_agents) while
carrying over the previous probe_failures count. Indivisible w.r.t. the
increment/reset scripts (Redis single-threads an EVAL).
#56 (notify to prompter/secretary refused) verified SAFE — the pin tests
(test_notify_rejects_prompter_recipient / _secretary_recipient /
_allows_ceo_recipient) already cover the only human notify target invariant;
no legitimate send is dropped, no code change.
* [chore] foundation/policy: spec gates + QA retry-key pin (Cluster F)
#50 sync_branch composes=() so the spec gate accepted a terminal/paused/
blocked task and the handler rebased a dead/parked branch — add a
PRECONDITION_SYNC_BRANCH_STATE (claimed/in_progress/verifying/needs_revision
only), rejection_kind=invalid_state. TDD: 28 spec tests.
#148 submit_root's prose asserts 'a Main-PM root is planning-typed, never
code' but only the creation path (main_pm_cannot_own_code) backed it — add
PRECONDITION_ROOT_NOT_CODE on the submit_root IntentSpec (defense in depth),
scoped to submit_root only so the shared submit_for_review action keeps
cell_pm+code submit_up parity. Graceful on Mock/None task_type so the
choreographer Mock-task tests don't crash. TDD: 2 spec tests.
#150 VERB_RETRY_LIMITS is keyed by the MCP-exposed names (pass/fail), not
the IntentSpec-internal pass_review/fail_review — already correct; add a
pin test so a one-sided rename can't silently drop the QA-handoff cap.
#142 main_pm_cannot_own_code/pm_cannot_own_code already normalize casing
(.lower()) — no-op, pin test test_main_pm_cannot_own_code_is_case_insensitive
already in tree.
* [chore] worksession-git: 405 merge-method fallback + non-destructive close (Cluster W)
#108 _merge_with_retry hardcoded 'squash' and raised MergeConflictError on a
405 with no method fallback — wedging the PM on an open, mergeable PR whose
repo merely had the squash button off. Add a 405 fallback to a permitted
method (via _first_allowed_merge_method, exclude='squash'), mirroring the CEO
merge_pull_request path. A 405 with no permitted fallback (or a second 405)
still falls through to the already-merged disambiguation / MergeConflictError.
TDD: 2 new tests (fallback-success, no-permitted-method-raises).
#109 close_pull_request defaulted delete_branch=True, so the choreographer
supersede path deleted a superseded PR's branch while the orchestrator
supersede path explicitly preserved it — the two disagreed, and the
destructive default ran on the 'close the dead PR' path where the branch may
still be referenced / useful for audit. Flip the default to False (opt-in
deletion) and make the choreographer caller explicit (parity with the
orchestrator). TDD: 1 new test (default preserves branch); existing
deletion-when-requested test now passes delete_branch=True explicitly.
Dispositions verified against current code (no silent drops):
- #27 REFUTED/FIXED-UNDEPLOYED: work_session.merge_pr resolves by session_id
(no global pr_number lookup); the real cross-repo collision fix
(project_id scoping on pr_merge/close_pull_request/rebase_pr_for_task/
pr_target) is already in tree + tested (test_pr_merge_scopes_task_lookup_
by_project_id, test_close_pull_request_scopes_task_lookup_by_project_id,
test_git_pr_target_scoping). Verify-only.
- #106 REFUTED: a guard exists (rev-list --count {base_ref}..{branch} == 0)
before reset --hard + base_ref falls back to default_branch; tests lock
the safety (test_create_branch_never_repoints_branch_with_real_work,
test_create_branch_does_not_reset_or_checkout_shared_clone).
- #218 BY-DESIGN: the merge_pr idempotent guard intentionally preserves the
audit trail (docstring + test_merge_pr_idempotent_on_already_completed_
preserves_audit_trail); a COMPLETED session always carries attribution
(COMPLETED only via merge_pr), so the NULL-COMPLETED case is unreachable.
- #104 BY-DESIGN: agents never merge to the repo default branch in RoboCo's
model (root→master is CEO-only); the guard is a correct CEO-only rail,
locked by test_pr_merge_into_default_branch_is_ceo_only.
* [chore] llm: surface disabled-provider downgrade + scrub probe log (#20/#3/#211)
#20/#3 resolve_for_agent silently fell through to the legacy Anthropic path
when a configured provider was disabled — indistinguishable from 'no
assignment', so the operator got no signal that spawns bypassed the
provider. Surface the bypass with a warning (graceful degradation stays the
default — a stalled spawn is worse than a routing miss) and add an opt-in
ROBOCO_ROUTING_STRICT (default-off) that fail-closes instead. Wired into the
panel Feature Flags card. TDD: 3 unit tests (warn-on-disabled, strict-raises,
no-assignment-stays-silent).
#211 probe_ollama_tags logged str(exc) raw on the generic-exception branch —
structured log could carry connection internals / stack traces. Log the
exception class name only. TDD: existing generic-branch test strengthened to
assert the log kwargs don't leak the raw text.
* [chore] support/stream/optimal/playbook/comms hardening (Cluster S)
Logical-gaps sweep, Cluster S (TDD, red→green per item):
#64 notification_delivery.acknowledge published the NOTIFICATION_ACKED bus
event directly (bypassing the outbox) — a rollback left a phantom ACK. Route
it through defer_bus_publish (after_commit), mirroring deliver.
#76 playbook.archive()/reject() stamped the archiver into approved_by/
approved_at, overwriting approval provenance (and fabricating approval for a
rejected draft). Add archived_by/archived_at (migration 053 + table + model)
and write those on archive/reject, leaving approval attribution intact.
#181 vector_store.replace_chunks wiped existing index rows even when every
chunk lacked an embedding (embedder failure). Skip the wipe when chunks is
non-empty but records is empty — preserve good rows for nothing.
#182/#183 optimal.record_learning recomputed a learn-{md5(full_content)}
tracking source that never matched the URI the plugin embedded chunks under
(roboco://learnings/{doc_id}, doc_id=lrn-{hash100}). Use the plugin's
returned doc_id so de-index/lookup-by-source finds the chunk rows.
#96/#97 transcription periodic flush only peeked ready buffers (unbounded
map growth) and ran sync callbacks on the event loop (a slow callback
blocked the flush task). Flush (remove) each ready buffer after notifying,
and offload each callback to a thread.
#212 _TEAM_SCOPED_ROLES was duplicated across communications/agents_config/
seeds. Single-source it in foundation.policy.communications; consumers
reference that object (identity-tested).
#19 stream_bus._dispatch_event re-ran already-succeeded handlers on a
recover_pending replay (duplicate side effects). Add a per-(event.id,
handler) SET-NX idempotency guard: skip on a hit, clear the key on handler
failure so a replay re-runs it, fail-open when redis is unavailable.
Dispositions (no code change): #77 approve() index-write pair asserted
BY-DESIGN; #62/#63 notification DB-dedup verified pinned; #184/#185 REFUTED;
#214 REFUTED; #215 BY-DESIGN.
* [chore] db/migrations: graph-integrity guard + conftest unreachable-DB warning (Cluster D)
Logical-gaps sweep, Cluster D (TDD + real alembic upgrade head verification):
#16/#37 add tests/unit/test_migration_graph_integrity.py — a static guard that
the alembic migration graph has exactly one head, every down_revision resolves,
every revision is reachable from a root, and no revision id is duplicated. The
suite builds its DB via Base.metadata.create_all (not alembic upgrade head), so
a forked head / dangling down_revision / duplicate id would otherwise ship
silently and break a real deploy mid-stream.
Caught a real bug in the process: migration 053's revision id
"053_playbook_archived_attribution" (33 chars) exceeded alembic's
alembic_version.version_num VARCHAR(32) — a fresh `alembic upgrade head` raised
"value too long for type character varying(32)" at the 053 stamp. Renamed to
"053_playbook_archived_attr" (26 chars). Verified end-to-end on a scratch PG:
upgrade head stamps 053, downgrade -1 returns to 052. (The pre-existing
test_every_migration_revision_id_fits_the_alembic_version_column guard is now
green too; it had been red on the 33-char id.)
#90 conftest silently pytest.skip'd every DB test when Postgres was unreachable
— a non-Docker box reported a green run of all-skips. Extract the warning into
_warn_if_pg_unavailable and fire it at import so the operator sees the DB is
down (the per-test skip path is unchanged). Test: warns when unavailable, silent
when reachable (verified under -W error::UserWarning).
Dispositions (verified against real code + a fresh alembic upgrade head, no code
change): #6 REFUTED — sa.Enum(create_type=False) at 001:119/304 does NOT break a
fresh upgrade head (001→052 applied cleanly on a scratch DB); #8 REFUTED — the
upgrade passed 030/031 (RAG chunk tables) without pgvector installed; pgvector is
a runtime concern handled by roboco/db/base.py, not a migration prerequisite;
#40 REFUTED — the `|| echo` mask was already removed and partial-schema drift
reports exit 1 (only by-design unreachable/unmigrated skips remain); #204 REFUTED
— the property walk seed IS pinned (random.Random(20260504), line 97); #205/#206
REFUTED — the smoke-trace fixture IS wired via
test_lifecycle_smoke_replay.py (8 passed); no shell smoke scripts exist in the
tree to wire; #137 BY-DESIGN — pyproject version 0.14.0 is an operational note,
no code gate.
* [chore] panel: admin-override force flag + kanban subtask_count + ws cleanup + ui-store dedupe (Cluster P)
#13: kanban admin-override into a hatch state (completed / awaiting_qa /
awaiting_pm_review) now requires an explicit force=true from the panel and
emits a dedicated task.admin_override audit row server-side; non-hatch
overrides need no force. Backend gate in tasks route + admin_set_status;
panel kanban-board sends force for hatch targets; TaskUpdate carries force.
#198: kanban service threads the real subtask_count (one grouped query) into
dev + priority-swimlane + main-pm-flat boards instead of a hardcoded 0.
#79: useWebSocket cleanup clears messages/lastMessage/state on unmount or
endpoint change so a dep-change (navigating to another stream) can't leak the
prior subscription's stale snapshot as live.
#186: disambiguate the duplicate ui-store modules -- the session/scroll store
in lib/stores renamed to useScrollRestorationStore / scroll-restoration-store
(barrel + 2 consumers updated); the sidebar/theme useUIStore in @/store is now
the sole useUIStore.
#12: verified already in-tree (release-proposal-card surfaces non-404 errors
with retry; getProposal maps only 404->null). #80 by-design (handleTransportError
already resets isSending on a no-payload SSE drop). #81 docs (streamUrl docstring
records that live-intake SSE auth is session-id-based bearer-style).
Backend: ruff+mypy clean, 278 tests green. Panel: lint+typecheck clean, 159 tests.
* orchestrator: park/reaper/readopt/a2a hardening + self-heal/ci-watch dedupe (Cluster O)
Closes the orchestrator-side logical gaps from the sweep:
- #75 a2a human-only drop surfaced: _dispatch_a2a_work logs the skip
("a2a request targets a human-only role; left as a notification for the
human (not spawned)") instead of silently dropping the target — the
CEO/secretary/prompter still see the notification; only the spawn is
suppressed. (orchestrator.py)
- #72 readopt liveness: _readopt_running_agents requires a non-stale live
claim (via _agent_holds_live_claim) and skips a zombie container so a
reaped-but-restart-readopted agent isn't double-counted as active.
- #74 shutdown drain: stop() calls _flush_respawn_tracker so the durable
respawn counter write-throughs aren't lost on a clean stop.
- #71 resolve_wait active-guard + deferred liveness: a rate_limit_lifted
WaitingRecord is only confirmed-live after a _confirm_resume_liveness
probe (deferred deletion _resume_confirm_delay=30.0), and an
already-active agent short-circuits the repark. Scoped to
rate_limit_lifted records (the only ones at risk of a false lift).
- #73 stuck-Claude kill: _maybe_kill_stuck_claude + _claude_stuck_kill_ttl
(config.claude_stuck_kill_seconds) — a live container whose heartbeat is
stale past the grace AND whose gateway probe is broken is killed+evicted,
not protected forever by the reaper's live-skip.
- #230 verified FIXED-UNDEPLOYED: _gateway_broken_past_grace already
requires N consecutive false-broken probes (not one flaky streak); no
change, test added to pin the N-consecutive invariant.
- #43 self-heal per-observation dedupe: a fingerprint collapses repeat
CEO notifications for the same CI regression.
- #44 ci_watch dedupe by (git_url, workflow): a monorepo's multiple
workflows each get their own fix task (was collapsed by git_url alone).
- #49 identity.role_for_slug_or_none None-hardening: a stale/malformed
slug resolves to None and the human-only skip falls through to the safe
"not spawnable" path instead of crashing.
- #193 strategy engine: notify the CEO on a persistent assess failure
instead of failing silently in the background loop.
TDD: test_no_spawn_human_roles (a2a skip surfaced), test_orchestrator_
shutdown_drain (#74), test_provider_overload_break (#71), test_readopt_
running_agents (#72), test_resolve_wait_repark (#71), test_stale_claim_
reaper (#73/#230), test_strategy_engine_loop (#193, new),
test_self_heal_engine (#43), test_ci_watch_engine (#44), test_identity
(#49). All red->green.
* chore: make-quality green — xenon complexity refactors + mypy test fixes + lifecycle regen
No behavior changes. Brings the tree to a fully green `make quality` (the
base branch never passed the xenon B-rank gate on several blocks; the
lifecycle artifacts had drifted from the committed ceo_reject_to_pool edge).
Xenon B-rank refactors (extract a helper; preserve semantics exactly):
- api/routes/tasks.py: _apply_forced_status_override + _StatusOverride
dataclass bundle (update_task override block).
- services/task.py: _enforce_no_pm_code_on_create (create guards) +
_escalation_diverts_to_pool (collapses the two board/advisory +
main_pm+code divert branches into one predicate).
- services/prompter.py: _coerce_pm_code_to_planning (create_task_from_draft).
- services/notification.py: _duplicate_unacked_exists (_create_notification
purpose-based dedup query + ACK_REQUIRED_BY_TYPE gate).
- services/sequencing.py: _same_assignee_lane_edges (the undeclared-surface
same-assignee lane fallback at the tail of dev_task_collision_edges).
- gateway/choreographer/_impl.py: _pm_task_type_error static helper
(_validate_assignee_task_type compound PM guard).
- gateway/choreographer/pr_gate.py: _gate_review_event_verdict +
_gate_review_body static helpers (_post_gate_review_to_pr).
mypy test fixes (no type:ignore — banned; use typing.cast with quoted
strings per TC006):
- test_task_update_completeness: TaskUpdate(acceptance_criteria=None).
- test_bus: cast("Redis", _FakeRedis()); Redis import under TYPE_CHECKING.
- test_pr_merge_concurrency: capture AsyncMocks into locals before asserting.
- test_notification_delivery_phantom: cast("UUID", to_agents[0]).
Lifecycle artifact regen (owed from Cluster T #100 — the
awaiting_ceo_approval -> pending `ceo_reject_to_pool` edge was added to the
spec in
|
||
|
|
15effce014 |
Chore: 141 Gaps fill-in (#283)
* Updated uv.lock
* Bunch of fixes we need to verify first..
* feat(megatask): per-cell project map root-subtasks (multi-project, multi-cell)
A MegaTask root-subtask can now target an ad-hoc per-cell project map — a
third targeting shape that mirrors the existing product fan-out root. In
RoboCo a project is per-cell (ProjectTable.assigned_cell); a monorepo is N
per-cell projects sharing one git_url. So 'multi-cell' IS 'multi-project',
and a task may mix per-cell projects across products or include OSS-library
projects not in any product.
Storage: migration 052 adds task_cell_projects (mirrors product_projects;
unique per (task, team)). TaskTable gains a cascade-delete cell_projects
relationship; TaskCreateRequest / TaskCreate / Task response carry the map.
Policy: batch.is_branchless_coordination + is_valid_batch_shape gain a
has_cell_projects param — a root-subtask targets exactly one of project /
product / cell-map; the umbrella still targets none. TaskService passes
has_cell_projects at every predicate call site and persists the rows in
create(). _ensure_branch_for_task cuts feature/main_pm/{root} per distinct
project in the map (via _distinct_projects_for_task); _require_target_or_umbrella
and _validate_batch_membership accept the map shape.
Fan-out: every distinct_project_ids site (task.py branch creation, routes
_project_for_complete + _resolve_project_for_merge, orchestrator
_ambient_projects_for_task, pr_review._project_slug_for, git._project_for_task)
generalizes to first-distinct-project-of-map-or-product. Choreographer
_resolve_subtask_project resolves a delegated subtask's cell from the parent's
cell map. The product-scoped _slugs_for_product intake helper is unchanged.
Intake: prompter._draft_cell_map extracts the per-cell map from the_work[].
_validate_batch_scope counts distinct projects across all drafts' cells
(>=2 min stays; one 2-cell draft satisfies it). create_task_from_draft
persists cell_projects for >=2-cell drafts (project_id/product_id None),
collapses a 1-cell map to the single-project shape, and leaves single-cell
top-level project_id drafts unchanged. _resolve_owning_team routes a
multi-cell map to Main PM (coordination root, like a product root — a cell
PM can't delegate cross-cell). propose_draft/propose_batch tool descriptions
declare the per-cell project_id (both Claude SDK + grok runtimes).
The umbrella stays branchless / pure-coordination / submit_root-rejected;
the CEO-escalation pr_number gate is not widened (the map root is
is_umbrella=False, mirroring a product root, so submit_root supplies it).
Single-cell root-subtasks and everything below them are byte-for-byte
unchanged. Un-run MegaTask waves (multi-cell drafts) become runnable.
* [feature] Panel per-cell project picker + pnpm format infra
MegaTask root-subtasks can fan out across cells (be+fe, fe+uxui). Since a
RoboCo project is per-cell (ProjectTable.assigned_cell), a monorepo is N
per-cell projects sharing one git_url — so multi-cell IS multi-project. The
batch-review card now shows one project Select per the_work entry, scoped to
that cell's repos, instead of one Select bound to a single top-level
project_id. confirmBatch validates each cell's project is in scope and the
batch still spans >=2 distinct projects.
- prompter.ts: CellWork gains optional project_id (the per-cell picker seam).
- batch-review-card.tsx: per-cell Selects (one per the_work entry), scoped to
the cell's projects; legacy single-cell drafts keep the one-Select path.
- use-prompter.ts: updateBatchDraftProject edits per-cell (entryIndex); confirmBatch validates every cell; batchFromEvent parses per-cell map.
Also adds the missing pnpm format infrastructure (the panel had no formatter
at all): prettier devDep + .prettierrc.json (default-style config: 80-col,
double-quote, semi, trailing-comma-all) + .prettierignore, plus format /
format:check scripts. Only the 3 changed files above were reformatted; the
~222 pre-existing non-compliant files are left untouched (a wholesale reformat
is a separate explicit decision, not bundled into this feature).
* [fix] MegaTask verification: migration 052 enum + async cell-map read
Two real bugs surfaced running the full gate against a containerized
Postgres (and the orchestrator boot log):
1. Migration 052 crashed a real orchestrator boot with
'type "team" already exists'. The generic sa.Enum(create_type=False)
does NOT set the postgres enum's create_type attribute, so op.create_table
(checkfirst=False) emitted a redundant CREATE TYPE against the pre-existing
team enum. Switched to postgresql.ENUM(create_type=False) — the postgres-
native enum whose create_type _check_for_name_in_memos actually reads, so
the CREATE TYPE is suppressed. Verified: 051->052 upgrade against a DB where
the team enum pre-existed (the exact path that crashed) now succeeds;
downgrade 052->051 drops the table and preserves the shared enum; fresh
upgrade head clean. (Migration 016 has the same latent sa.Enum pattern but
never re-runs in prod, so it's noted, not touched here.)
2. _ensure_branch_for_task read task.cell_projects (lazy=selectin to-many)
directly, tripping MissingGreenlet on a freshly-created/unqueried task —
which then poisoned the async session (PendingRollbackError). Replaced with
_task_has_cell_map: peeks InstanceState.unloaded (no IO) and reads the
already-loaded map, falling back to an awaited count query only when the
relationship is genuinely unloaded. Non-ORM stubs route to the plain
attribute. Fixes 2 integration tests; the 6 cell-map unit tests still pass.
Also: typed the self stub as Any in test_choreographer_subtask_project
(mypy tests/ wants Choreographer, not SimpleNamespace) — the codebase idiom.
Gate: ruff format/check clean; mypy roboco/ + tests/ clean; full pytest
10371 passed / 388 skipped against containerized pgvector:pg16; vulture clean.
Pre-existing xenon C-rank on reassign (from prior commit
|
||
|
|
fa3e25e656 |
feat(grok): pluggable agent providers + Grok on the official grok CLI (#218)
* feat(providers): pluggable agent providers + Grok (xAI) backend
Add a roboco/llm/providers/ seam — an AgentProvider lifecycle ABC and a
ProviderRegistry keyed by ModelProvider — so the orchestrator can drive
agent backends other than Claude Code.
The first non-Claude backend is GrokProvider for xAI's grok-build-0.1.
xAI is OpenAI-compatible only (no Anthropic-Messages endpoint), so a Grok
agent runs an OpenAI-protocol runtime pointed at https://api.x.ai/v1
rather than the ANTHROPIC_BASE_URL injection the other providers use. It
reuses the orchestrator's existing mount/auth assembly, so it inherits the
same MCP gateway + tool-manifest wiring as every other agent by
construction, and passes its prompt via env (never an argv positional).
The change is purely additive: only GROK routes through the registry;
Anthropic / Ollama Cloud / self-hosted spawns run the existing
_spawn_container path unchanged.
Includes:
- ModelProvider.GROK (migration 038) + a seeded Grok provider row
(migration 039) + a grok-build-0.1 catalog entry
- GET/PUT /api/providers/grok-key to store the xAI key (Fernet-encrypted,
reusing the existing provider-key machinery)
- ClaudeCodeProvider reference adapter over the current spawn
- unit tests for the registry, GrokProvider (gateway wiring, no
ANTHROPIC_* leak, prompt-injection safety, failure paths) and routing
The dedicated roboco-agent-grok image and the exact OpenAI-protocol CLI
invocation are the remaining piece to finalise with xAI.
* feat(providers): native Grok runtime — opencode image, config gen, panel key
Complete the native Grok (xAI) path so grok-build-0.1 runs as a real
RoboCo agent, not just the provider seam.
- roboco-agent-grok image (docker/agent-grok.Dockerfile): FROM agent-base
+ opencode (the OpenAI-protocol runtime). One image serves every role;
role behaviour comes from the mounted manifest / mcp-config / system
prompt, exactly as on the Claude path.
- Entrypoint renders opencode.json at spawn from the GrokProvider env
contract + the mounted Claude Code mcp-config.json
(roboco.llm.providers.opencode_config): translates RoboCo's gateway
servers (roboco-flow / roboco-do / ...) into opencode's mcp block,
declares the xAI OpenAI-compatible provider + model, and wires
permissions + instructions. Pure, unit-tested translation.
- Orchestrator registers GrokProvider with the registry-qualified image
(_qualify_agent_image) so it resolves in local and registry deploys.
- Compose (both files + the registry compose) gain an agent-grok-image
builder service.
- Panel: a Grok (xAI) API key card on the AI Providers page, plus the
grok ModelProvider value.
KNOWN PARITY GAP (opencode runtime): the bash-guard PAT-scrub and the
transcript-based usage/cost capture are Claude Code hooks and do not
transfer to opencode. bash permission is operator-tunable
(ROBOCO_GROK_BASH_PERMISSION) so a deployment can fail closed until a
security/usage-parity opencode plugin lands. That plugin and live E2E
validation are the remaining work to finalize with xAI.
* ci(release): build + publish the roboco-agent-grok image
Add roboco-agent-grok to the release workflow's image build/publish map so
registry deploys carry the Grok runtime image (parity with every other
agent image). Split from the feature commit because pushing a workflow
change requires a workflow-scoped token.
* fix(migration): commit the grok enum value before seeding (autocommit_block)
CI's "Apply database migrations" failed with asyncpg
UnsafeNewEnumValueUsageError: alembic runs the whole upgrade in a single
transaction, so migration 039's INSERT used 'grok' in the same transaction
that 038 added it — which Postgres forbids. Splitting into two migration
files did not help (one transaction spans both). Wrap the ALTER TYPE ADD
VALUE in op.get_context().autocommit_block() so the value commits before 039
(and any later migration) uses it. Still renders in offline --sql, so the
enum-migration-parity test is unaffected.
* feat(grok): price grok-build-0.1 + secret-scrub opencode plugin
- pricing.py: add grok-build-0.1 rates ($1/1M input, $0.20 cached, $2/1M
output), verified against xAI's published pricing. Grok is a priced
non-Anthropic model, so cost computes the moment usage is captured.
- secret-scrub.js: an opencode tool.execute.before plugin porting the
security-critical bash-guard deny rules (git network ops, credential-file
reads, /proc env, internal-host HTTP, roboco.* imports, ROBOCO_AGENT_ID
forgery, env dumps, destructive rm) to the opencode runtime — restoring the
guard the Claude Code hook can't provide there. Throwing denies the call
(confirmed by opencode's env-protection example). Wired into the generated
opencode.json plugin array + baked into the grok image.
Deny logic verified via node (9 deny + 5 allow cases). UNVALIDATED against a
live opencode runtime: confirm it fires in the live E2E spawn before a Grok
dev-agent touches a real repo; the bash permission is operator-tunable as a
second gate.
Cost CAPTURE (distinct from pricing) is intentionally NOT built yet: opencode's
plugin hooks expose model info but no token/usage object, so the capture path
is unconfirmed and needs the live spawn to settle.
* feat(grok): read opencode session usage for cost capture
Confirmed by inspecting a local opencode run: opencode persists per-session
usage in SQLite at ~/.local/share/opencode/opencode.db — the `session` table
carries cost + tokens_input/output/reasoning/cache_read/cache_write. xAI's
response usage object (prompt_tokens, completion_tokens,
prompt_tokens_details.cached_tokens, completion_tokens_details.reasoning_tokens)
maps directly onto those columns.
Add opencode_usage.read_session_usage / cost_for_session: read the opencode DB
and price the tokens via roboco.billing.pricing (our cost stays authoritative;
opencode's own `cost` column is kept for reference). Tested against a fixture DB
mirroring the real schema (single session, summed sessions, missing/empty DB).
Remaining wiring (for the live spawn): mount the opencode data dir on grok
spawn + call cost_for_session at reap to record the usage rollup.
* fix(grok): correct opencode provider (Responses API), stdin, reasoning cost
A live opencode run against api.x.ai/v1 surfaced three real bugs:
1. Provider package — grok-build-0.1 is driven via the OpenAI Responses API
(opencode calls model.responses()). @ai-sdk/openai-compatible is
chat/completions only and errors "responses is not a function". Switch the
generated opencode.json provider + the grok image to @ai-sdk/openai.
2. Headless hang — `opencode run` blocks after init without a TTY; close stdin
(`< /dev/null`) in the entrypoint so it proceeds to the model call.
3. Reasoning-token cost — grok-build-0.1 is a reasoning model; reasoning tokens
bill as output but opencode stores them in a separate column. cost_for_session
folds tokens_reasoning into output (else ~22x undercount).
Verified end-to-end against a real session row (input=6120, output=1,
reasoning=226, cache_read=1856): our pricing reproduces opencode's stored USD
cost ($0.0069452) exactly. Tests anchored to that real row.
* feat(grok): first-class xAI/Grok routing mode (UI + backend)
The Routing-mode toggle had Anthropic / Ollama / Self-Hosted / Mix but no way
to route the whole org to Grok. Add it end to end:
- backend: apply_mode("grok") + _apply_grok (GLOBAL default -> grok-build-0.1) +
derive_mode "grok" detection; ApplyModeRequest/ModeResponse accept "grok".
- panel: a "Grok" routing-mode card (between Anthropic and Ollama, gated on the
xAI key) + flipToGrok; a Grok group in the per-agent mix dropdown +
catalogGrokOnly + a grok ProviderBadge variant; the mix-save key check and
the AI-routing description now cover Grok.
- tests: integration derive_mode/apply_mode "grok" cases (+ grok provider row
in the fixture).
Gated: ruff + mypy clean; panel typecheck + lint clean.
* feat(grok): reasoning-effort by role (cut grok-build cost on cheap roles)
grok-build-0.1 reasons heavily by default and reasoning bills at the output
rate (a live "say ok" call emitted ~300 reasoning tokens, ~85% of its cost).
Confirmed live that opencode's `--variant minimal` cuts reasoning ~54%
(298 -> 136 tokens, same prompt).
GrokProvider now picks reasoning effort by role: code-quality roles (developer,
qa, pr_reviewer) keep full reasoning; coordination / docs / board roles
(cell_pm, main_pm, documenter, product_owner, head_marketing, auditor, prompter,
secretary) run "minimal". It's passed to opencode via the entrypoint's
`--variant`. Operators can force one effort for ALL grok agents with the
ROBOCO_GROK_REASONING_EFFORT env (minimal | high | max, or default/full).
Tests cover the role map, the env override, and the spawn env wiring.
* style(panel): show the Grok (xAI) key card above the Ollama card
* fix(grok): stop opencode subagent-stream hang at the config layer
The Grok pr_reviewer wedged in_progress forever: opencode's default agent ran
with the subagent `task` tool enabled, spawned an Explore subagent on
grok-build-0.1 whose model call opened an SSE stream that went idle, and the
run hung with no timeout.
- Hard-disable opencode's subagent `task` tool in the generated opencode.json.
No RoboCo role uses opencode-internal subagents — work flows through the
gateway verbs — so removing the tool kills the hang trigger outright.
- Set provider.xai.options.timeout + chunkTimeout (operator-tunable via
ROBOCO_GROK_REQUEST_TIMEOUT_MS / ROBOCO_GROK_CHUNK_TIMEOUT_MS) as the
defence-in-depth backstop; chunkTimeout aborts an idle stream.
- Bundle the permission + timeout + subagent knobs into an OpencodeGuards
dataclass (keeps the builder under the arg-count gate).
- Drop the dead ROBOCO_AGENT_TOOLS spawn env (it had no consumer); opencode
tool restriction lives in the rendered config now.
* feat(grok): reaper watchdog kills wedged opencode containers
The heartbeat reaper deliberately skips a task whose assignee holds a live
ACTIVE container, so a Claude agent deep in a long edit/test cycle isn't
churned out from under live work. A wedged opencode container breaks that
assumption: it stays ACTIVE while firing no gateway verb, so its heartbeat
never advances and the live-instance skip would shield its task forever — the
exact way the Grok pr_reviewer parked in_progress.
Add a longer grok-idle kill threshold (ROBOCO_GROK_IDLE_KILL_SECONDS, default
900s, well past the stream chunk timeout). A GROK instance idle past it is
force-removed (its logs dumped to disk first) and evicted from the instance
registry, so the same reaper pass then releases the task. Only GROK runtimes
are eligible — a quiet Claude agent keeps the heartbeat-skip protection.
* feat(grok): guard interactive roles from GROK routes (interim)
intake (prompter) and secretary run a held-open chat session driven by the
Claude Agent SDK. GROK has no interactive runtime yet, and a GROK route for
those slugs would be spawned with the route creds injected as ANTHROPIC_*
against api.x.ai/v1 — the wrong protocol — producing a silent, empty reply
(the blank intake we observed).
Downgrade a GROK route for intake-1/secretary-1 to the Anthropic default with
a logged warning. The one-shot delivery roles route to GROK unchanged. This
guard is replaced by the real interactive fork once the opencode interactive
driver lands.
* feat(grok): capture one-shot Grok usage/cost from the opencode store
A GROK agent runs opencode, not Claude Code: it has no SDK /usage/status
server and writes no Claude transcript, so _resolve_final_token_usage found
nothing and every Grok agent finalized at 0 tokens / $0 — the opencode_usage
reader existed but had no caller.
- Mount a per-agent opencode data dir ($DATA/opencode/<agent_id> →
/home/agent/.local/share/opencode) so opencode.db is captured, and mount the
same host dir into the orchestrator (/data/opencode) in all three compose
files so the finalizer can read it back — the opencode analogue of the
mounted Claude transcript.
- _resolve_final_token_usage branches on provider_type: GROK reads opencode.db
via opencode_usage (reasoning folded into output, billed at the output rate)
and skips the SDK/transcript path. A 0-token read logs a WARNING so a silent
mount failure isn't mistaken for a real zero-cost run.
- ROBOCO_OPENCODE_DATA_DIR overrides the in-orchestrator path for local runs.
* feat(grok): make interactive spawns first-class on AgentProvider (additive)
The AgentProvider ABC modelled only the one-shot lifecycle (spawn/stop/
health_check/remove), so the interactive intake/secretary roles could never
route through a provider. Add an opt-in interactive surface:
- supports_interactive class flag (default False).
- InteractiveSpawnSpec: the resolved AgentConfig + session id + role-specific
image + optional HMAC token — everything a provider needs without importing
orchestrator internals.
- spawn_interactive(spec): a non-abstract default that declines via
ProviderError, so every existing one-shot provider is unchanged.
Pure scaffolding — no provider opts in yet (GrokProvider flips the flag when
its interactive driver lands). Zero behavioural change.
* feat(grok): Grok-native interactive runtime (opencode serve) — container side
Builds the Grok analogue of the Claude intake/secretary live-session runtime,
satisfying the same IntakeSession seam so the existing IntakeDriver loop,
message source, relay, and StreamChunk panel contract are reused unchanged:
- OpencodeServeSession: a held-open `opencode serve` session (context persists
across turns) where each human turn is one synchronous POST /session/:id/
message; normalize_opencode_message maps the reply parts to text/thinking/
tool_use/draft/turn_end chunks (draft via a propose_draft tool part or the
fenced roboco-draft fallback). Doc-verified against opencode's server API.
- grok_intake_main / grok_secretary_main: container entrypoints mirroring the
Claude mains but yielding an OpencodeServeSession; they render opencode.json
(xAI provider + MCP + system prompt) first, then run the receiver + driver.
- roboco-agent-grok-prompter / -secretary images (FROM roboco-agent-grok) +
their builder services in all three compose files.
UNVERIFIED-LIVE: the opencode serve flow + exact Part schema + draft path need
a live run against grok-build-0.1 (the part mapping is defensive). The
orchestrator wiring that routes a GROK intake/secretary route to these images
is the next step (a design decision is open — see the handoff notes).
* feat(grok): route interactive intake/secretary to opencode-serve images
Wire the GROK interactive path the in-place way (matching how the interactive
roles already choose ANTHROPIC_* per route), so a GROK route launches the
Grok-native opencode-serve image instead of the Claude SDK-driver image:
- _spawn_intake_container / _spawn_secretary_container pick the
grok-prompter / grok-secretary image (ensuring the base→grok→interactive
build chain) when the route is GROK, and stamp provider_type on the spec +
AgentConfig so finalize routes usage to the opencode store.
- _build_intake_run_cmd / _build_secretary_run_cmd inject OPENAI_* + the
opencode store mount + system-prompt env for GROK via a shared
_append_interactive_provider_env, keeping ANTHROPIC_* for every other
provider. The intake's minimal mounts (no gateway MCP) are preserved, so
Grok intake matches the Claude intake's tool surface (the spec).
- Add a per-agent opencode store mount to the interactive host paths so
interactive Grok usage/cost is captured like the one-shot path.
Removes the interim Phase-0 routing guard (the real path supersedes it) and
retires the unused AgentProvider.spawn_interactive/InteractiveSpawnSpec seam —
the interactive roles have a bespoke assembly that the one-shot provider
surface doesn't fit, so the fork lives in their own builders.
UNVERIFIED-LIVE: end-to-end intake/secretary chat on Grok needs the stack up +
opencode serve confirmed against grok-build-0.1.
* feat(grok): surface intake/secretary in the mix-mode picker; doc guardrail parity
- Panel: add intake-1 (prompter) and secretary-1 to the mix-mode per-agent
routing list so an operator can assign Grok (or Claude) to the interactive
roles from the UI; assigning a Grok model routes them to the opencode-serve
image. tsc + eslint clean.
- opencode_config: correct the now-stale parity note — bash-guard is ported
(secret-scrub.js) and usage/cost is captured (opencode store); the remaining
gap is the budget/loop/stop/prompt-injection hooks, which need a sidecar
plugin (open decision), with ROBOCO_GROK_BASH_PERMISSION as the interim gate.
* test(grok): mypy-clean the reaper watchdog + interactive spawn tests
The CI mypy scope (roboco/ tests/) flagged test-only typing issues my per-file
runs missed: direct method assignment (orch._remove_container = AsyncMock())
trips [method-assign], and a module-level dict[str,str] is invariant against
the dict[str, str|None] the run-spec expects.
- Use monkeypatch.setattr for _remove_container in the watchdog tests.
- Annotate the shared _HOSTS as dict[str, str | None].
Production code unchanged; mypy roboco/ tests/ is green.
* feat(grok): cost-ceiling kill-switch (budget-guardrail parity)
Claude Code's per-agent token-budget hook fires against the SDK :9000 server;
opencode exposes NO usage/budget hook to a plugin (confirmed against its plugin
docs), so the budget kill-switch can't be a plugin/sidecar — the orchestrator
enforces it instead.
_enforce_grok_cost_budget runs each dispatch tick: for every ACTIVE GROK
container it reads cumulative cost from the opencode store (the Phase-2 reader)
and kills + evicts it past ROBOCO_GROK_MAX_COST_USD (0 = off), after which the
reaper releases the freed task. This also catches a runaway loop that keeps
firing verbs (so it evades the idle watchdog) but still burns cost.
Covers the budget/runaway-burn slice of guardrail parity. The remaining Claude
hooks (prompt-injection PRE-gate, stop-guard terminal-verb) have no blocking
opencode equivalent — opencode's message/stop hooks are observe-only — and the
interactive reasoning-variant has no opencode.json/serve knob (CLI-flag only);
both are pinned for a live probe rather than shipped as a guess.
* docs(grok): document Grok's reduced guardrail posture (honest, not blocking)
Grok agents run on opencode, not Claude Code, so they do NOT have full
guardrail parity — claiming otherwise would be false. Document it truthfully
and keep them usable rather than blocking them.
- Panel routing card: an amber caveat shown in Grok/Mix mode — command/
secret-exfil guard + cost cap apply to Grok, but the prompt-injection guard
does NOT (opencode cannot block a turn); Anthropic/Ollama/Self-Hosted run
through Claude Code with the full guard set; prefer those for agents that
ingest untrusted or cross-agent content; Grok is safe for trusted work.
- docs/self/architecture/llm-provider-security.md: the reference — the two
runtimes, which provider uses which, the per-guardrail parity matrix, why
the injection/stop gaps exist (opencode hooks are observe-only), and the
routing recommendation (delivery roles handling untrusted content → a
Claude-Code-runtime provider).
Panel tsc + eslint clean.
* fix(grok): make the live interactive path work — store perms, error surfacing, variant
Found by actually running opencode serve locally (the path was doc-verified but
never executed). Three fixes:
1. EACCES on the opencode store mount (the live intake crash): on Linux docker
auto-creates a missing bind source as root:root, so the non-root agent user
could not mkdir/write in /home/agent/.local/share/opencode and opencode died
at boot. _ensure_opencode_data_dir pre-creates the per-agent dir 0777 before
the mount (one-shot via the _GrokHost seam, interactive in both spawns).
2. Silent blank reply on a model error: opencode reports a turn failure in
info.error with parts=[], NOT as a part — verified live (a bad xAI key
returns info.error APIError). send() / normalize_opencode_message now surface
it as an "error" StreamChunk so a failed turn is never blank (the original
intake bug class). Confirmed live: the error now renders.
3. Reasoning variant on the serve path: the live OpenAPI shows the message body
accepts a "variant" field (it is NOT CLI-only, as the docs implied), so the
pin is unblocked. send() passes ROBOCO_GROK_VARIANT as the per-turn variant;
the orchestrator sets it per-role (_reasoning_effort_for) for interactive
Grok, the same lever as the one-shot --variant.
opencode serve startup, POST /session, session-id extraction, the part-type
mapping (text/reasoning/tool), and the error path are all validated against a
live opencode 1.17.8. A real successful grok reply still needs a funded key.
* fix(grok): pre-create agent-owned ~/.local in the grok image (opencode state EACCES)
Running the built grok-prompter container surfaced a second EACCES the
mechanism analysis missed: bind-mounting the opencode store at
~/.local/share/opencode makes docker create the intermediate ~/.local AS ROOT,
so the non-root agent user then cannot create its sibling ~/.local/state and
opencode dies at boot. Pre-create the ~/.local tree agent-owned in the image so
the mount leaves the parents writable. Complements the orchestrator 0777
host-source pre-create (which covers the bind source on Linux).
Verified live: with this fix the container starts clean, opencode serve opens
the session, a POST /turn produces a real grok reply, and all chunks
(thinking/text/turn_end) reach the relay endpoint.
* feat(grok): prompt-injection guard for Grok (parity with the Claude hook)
The injection guard is RoboCo's own hook (user-prompt-hook.sh), not a runtime
built-in, so it can be recreated at our input boundary regardless of runtime —
opencode's lack of a blocking pre-prompt hook is irrelevant.
- prompt_guard.detect_injection: the deny patterns ported to reusable Python.
- IntakeDriver._run_turn scans every interactive turn before sending it to the
model and denies a match as an error chunk. Covers BOTH Grok (opencode) and
the Claude SDK intake (which runs with setting_sources=[] and so never loaded
the bash hook — it was unguarded too).
- The one-shot grok entrypoint scans ROBOCO_INITIAL_PROMPT and refuses a
poisoned task prompt (parity with the Claude UserPromptSubmit deny).
- Broadened the pattern (Python + the bash hook, kept in sync) to catch the
multi-qualifier canonical phrasing "ignore all previous instructions", which
the single-qualifier original missed — without false-positiving on
"ignore the linting rules" (an intermediate non-qualifier word breaks it).
So Grok now has the command/secret-exfil guard (secret-scrub), the cost cap,
AND the injection guard. Verified: 94 agent_sdk tests pass; bash + Python agree
on detect/miss cases.
* docs(grok): drop the security disclaimers — injection guard closes the gap
With the prompt-injection guard now recreated for Grok (prior commit), the
"Grok lacks the injection guard / prefer Claude for delivery roles" warning is
no longer true, so remove it:
- Panel routing card: replace the amber "prefer Claude / not safe" caveat with
a neutral one-liner — Grok agents run on opencode; the command/secret-exfil
guard, the prompt-injection guard, and the cost cap all apply.
- docs/self/architecture/llm-provider-security.md: prompt-injection row flips to
"yes" for Grok; intro + routing recommendation updated to "effective security
parity, any agent (incl. delivery roles) can run on Grok"; the only remaining
unported hook is the non-security stop-guard.
- opencode_config docstring: the remaining gap is now just the stop-guard
(budget + injection are covered).
Panel tsc + eslint clean.
* fix(grok): allow external-directory reads so the pr-reviewer can work
Live NAS run showed the Grok pr-reviewer claim the review and fetch the diff,
then write it to /tmp and FAIL to read it back: opencode auto-denied
"external_directory (/tmp/*)" — its file tools refuse paths outside the project
cwd, and in headless serve/run mode an "ask" permission auto-rejects (no human).
Add permission.external_directory (default "allow", env
ROBOCO_GROK_EXTERNAL_DIR_PERMISSION) to the generated opencode.json. The
container is the sandbox and secret-scrub still blocks credential-file reads, so
allowing in-container external-dir reads is safe and unblocks legitimate scratch
use (e.g. the pr-reviewer grepping a large diff in /tmp).
Verified live against grok-build-0.1: with external_directory:"allow" the Read
tool reads a file outside cwd and returns its contents (no auto-reject); the
plain-string form is accepted by opencode 1.17.8.
Needs a rebuild of roboco-agent-grok + a pr-reviewer re-run on the NAS to confirm.
* refactor(grok): split eligibility out of _maybe_kill_wedged_grok (xenon C -> B)
CI complexity gate (make quality -> xenon --max-absolute B) flagged
_maybe_kill_wedged_grok at rank C — too many guard branches in one method.
Extract the kill-candidate decision into _wedged_grok_slug(task, last_heartbeat)
-> slug | None (recent-heartbeat / no-owner / not-ACTIVE / not-GROK all yield
None); _maybe_kill_wedged_grok now just kills + evicts the returned slug.
Behaviour is identical (same guards, same order) — the reaper watchdog tests
pass unchanged. xenon now passes on the full package; ruff + mypy clean.
* feat(grok): start the in-container SDK server + budget feed (Claude parity)
The keystone of the Grok parity work (CEO's "take Claude as baseline, create
what's missing" call): the one-shot Grok container now starts the same SDK
server the Claude path runs, so the per-verb circuit breaker (the flow/do MCP
servers already POST /verb/attempted to it), the per-session budget/loop
counters, the terminal-verb tracking, and the SessionEnd post-mortem all work
on Grok instead of being silently absent.
- entrypoint: launch roboco.agent_sdk.server (bare venv python, not `uv run`
which would re-sync the drifted clone lock and stall), wait for /health,
reset counters; run opencode WITHOUT exec so the script regains control to
run the post-mortem and the silent-exit substitute after the run returns.
- budget-feed.js: opencode plugin that gates on /budget/status in
tool.execute.before (halt/loop deny — the only place to stop a runaway
one-shot run; opencode has no PostToolUse-deny) and records the executed
tool + args-hash in tool.execute.after. Fail-open; bare-verb normalization
for MCP-namespaced terminal verbs.
- silent-exit substitute: on a graceful exit with no terminal verb the
entrypoint posts /terminal/force_substitute so the task isn't left stuck
claimed/in_progress (Stop-hook parity at the boundary).
- opencode_config: wire budget-feed into the plugin array; add
ROBOCO_OPENCODE_EXTRA_PLUGINS so per-image role tool plugins load scoped to
one role; read the per-role ROBOCO_GROK_EDIT_PERMISSION.
Targeted gate green (ruff/mypy/xenon + opencode_config tests; node --check on
the plugins; bash -n on the entrypoint).
* feat(grok): give the Grok Secretary its CEO-authority tools (blocker)
The Grok Secretary could chat but had zero directive tools — it could not read
company state or act on a CEO command, so it was non-functional. This is the
integration blocker.
- secretary-tools.js: opencode plugin registering read_company_state /
read_task / submit_directive via the Hooks.tool API, each calling
/api/secretary/* with the container's HMAC agent token — a direct port of the
Claude Secretary's SDK tools (secretary_driver.build_secretary_options). The
high-impact directive kinds stay gated server-side (queued for CEO confirm).
- agent-grok-secretary.Dockerfile: bake the plugin and scope it to this image
via ROBOCO_OPENCODE_EXTRA_PLUGINS, so only the Secretary carries CEO authority.
- grok_secretary_main: correct the docstring that falsely claimed the tools
reached the API "through the mounted MCP gateway" (there is no gateway mount;
they're an opencode plugin).
- secretary.md: name the three tools and restate the confirm-before-act gate.
Verified locally that opencode loads a file-path plugin importing
@opencode-ai/plugin and resolves the package; the live model-tool-call +
backend round-trip is flagged UNVERIFIED-LIVE for the NAS.
* feat(grok): give the Grok Intake its propose_draft tool (draft card)
The prompter prompt tells the model to call propose_draft when the spec is
ready, but on Grok that tool didn't exist — so no draft chunk, no panel draft
card, and the human couldn't launch a task from a Grok intake chat.
- intake-tools.js: opencode plugin registering propose_draft via Hooks.tool;
the execute() only ACKs — the driver (OpencodeServeSession.normalize ->
_is_propose_draft -> _draft_from_tool_input) intercepts the tool CALL and
emits the `draft` chunk the panel renders.
- agent-grok-prompter.Dockerfile: bake the plugin, scoped to this image via
ROBOCO_OPENCODE_EXTRA_PLUGINS (delivery roles never draft).
- test: a propose_draft tool part normalizes to a draft chunk (not a tool_use).
The live tool-call -> draft-card path is flagged UNVERIFIED-LIVE for the NAS.
* feat(grok): scope opencode edit/bash/external-dir permissions per role
Grok wrote ONE global permission block, so a Grok pr_reviewer (or qa / PM /
auditor) ran with edit=allow + bash=allow on untrusted PR content. Now the
permissions are derived per role, mirroring orchestrator._get_role_permissions
on the Claude path:
- edit — allow only roles that write code (role_config.allows_write:
developer / documenter); everyone else edit=deny.
- bash — allow only roles that legitimately run a shell (developer /
documenter / cell_pm / main_pm); the read-only reviewers (qa / pr_reviewer /
auditor) and the board get bash=deny. secret-scrub still guards the rest.
- external_directory — only the pr_reviewer reads scratch outside its cwd (the
/tmp diff); delivery roles get deny.
One-shot roles resolve these in GrokProvider._append_grok_env; the interactive
intake/secretary set edit=deny + bash=deny in the orchestrator (intake keeps
external-dir reads for sibling product repos, the secretary does not). The
Claude path is untouched — the permission env is a GROK-only contract.
Targeted gate green (ruff/mypy/xenon + provider + interactive-spawn tests).
* feat(grok): park the provider on an xAI 429 (break the respawn loop)
A one-shot grok run that hit an xAI 429 exited without a terminal verb; the
dispatcher then re-spawned the same task every tick (429 -> exit -> respawn), a
container/token/cost loop with no living agent to call i_am_blocked.
- entrypoint: detect a rate-limit signature in the run output and exit 75
(EX_TEMPFAIL); a rate-limited task is NOT substituted — it must be retried.
- _handle_stopped_container: on a grok exit 75, park the provider via the
rate-limit tracker (retry_after window) instead of crash-retrying, and don't
count it as a crash. The existing probe-resume loop clears the park after the
window (unknown-provider time-expiry fallback) and the task is retried.
- spawn_agent: a grok-only, fail-open guard skips the launch while the provider
is parked, so the dispatcher no-ops instead of looping. The Claude path is
untouched.
Targeted gate green (ruff/mypy/xenon + new rate-limit tests; bash -n on the
entrypoint).
* feat(grok): close the secret-scrub bash-guard parity gaps
secret-scrub.js (the opencode bash guard) was missing three rules the Claude
bash-guard hook has, leaving a Grok dev able to read secrets the Claude path
blocks:
- source / dot-source of a credential-bearing file (source .env, . ./.env,
.bashrc / .git-credentials / .netrc / /proc/*/environ).
- interpreter one-liner reading a credential file
(python -c "open('.env')", node -e "readFileSync('.git-credentials')").
- git-ops check now runs on a SKELETONIZED command (heredoc bodies + echo/printf
args stripped) so a README/heredoc that merely documents `git push` is no
longer mistaken for invoking it — a false-positive parity fix from the Claude
guard.
Functionally smoke-tested with node against the real plugin (git push denied;
echo/heredoc "git push" allowed; source/interpreter cred reads denied; normal
commands allowed). Live opencode firing stays flagged in the file header.
* fix(grok): record a usage session for interactive intake/secretary (M1+M7)
_spawn_intake_container / _spawn_secretary_container built the AgentInstance by
hand and never recorded an agent_spawn_sessions row, so the reap finalizer had
no usage_session_id to look up — every interactive session (Claude or Grok)
finalized at 0 tokens / $0 in the rollups. Record the session (task_id=None) and
pin its id on the instance, mirroring _launch_spawn; the GROK path reads
opencode.db by this id, the Claude path reads the transcript.
Also correct the grok_intake_main docstring (M7): it claimed the serve process
was "gateway-wired" with an "MCP gateway", but interactive intake mounts no
gateway — its only tool is propose_draft, registered by the intake-tools.js
plugin.
* fix(grok): surface a dead opencode-serve clearly instead of a zombie chat (M2)
If `opencode serve` died after the session opened, every subsequent turn failed
with an opaque httpx connection error while the container lingered. send() now
detects the exited subprocess (returncode set) and yields a clear error chunk +
turn_end so the panel shows a real "session ended — start a new chat" message;
the idle watchdog / a human reap then tears the container down.
* fix(grok): close the panel relay when the cost-cap kills an interactive chat (M4)
_enforce_grok_cost_budget killed + evicted a container directly. For the
interactive roles (intake/secretary) that left the panel SSE relay open with no
close sentinel, so the chat froze with no explanation. Add
PrompterLiveRegistry.close_by_agent (push a final error event, then close every
session bound to that agent) and call it from the cost-cap watchdog when the
killed agent is the intake or secretary, so the panel reports the chat ended on
the cost cap instead of hanging.
* fix(grok): make the opencode runtime actually load — proven live on grok-build-0.1
Live verification (opencode 1.17.8 + grok-build-0.1, funded key) showed the Grok
runtime was loading INERT, three ways:
1. The provider override `provider.xai.npm=@ai-sdk/openai` failed model
resolution (ProviderModelNotFoundError) — opencode can't resolve that package
from its module path. Worse, ANY custom `provider.xai` block (even just
options) breaks plugin-tool registration. opencode's BUILT-IN xai provider
drives grok-build-0.1 with working tool-calls, so emit NO provider block; the
key + base reach it via XAI_API_KEY / XAI_BASE_URL env (provider.options.apiKey
alone does NOT authenticate).
2. Plugins referenced by absolute path in the config `plugin:` array never
registered their hooks/tools. opencode 1.17.8 only registers from the plugin
AUTO-DISCOVERY dir (~/.config/opencode/plugin/). Bake all plugins there.
3. Plugins must use a NAMED export, not `export default`.
Changes:
- opencode_config: no `provider` block, no `plugin` array; drop the dead
XaiTarget + timeout machinery; build_opencode_config now takes a model string.
- GrokProvider / orchestrator interactive env: inject XAI_API_KEY + XAI_BASE_URL
(drop the now-unused OPENAI_*).
- secret-scrub / budget-feed / secretary-tools / intake-tools: named exports;
baked into /home/agent/.config/opencode/plugin/ (drop the EXTRA_PLUGINS env).
- agent-grok* Dockerfiles: plugin dir + agent ownership; drop the unneeded
@ai-sdk/openai global install.
Verified live end-to-end: grok-build-0.1 calls read_company_state AND
submit_directive through secretary-tools.js and the backend receives both with
the agent token; a tool.execute.before guard fires; built-in tool-calls work.
Targeted gate green (ruff/mypy/xenon + opencode_config/providers/interactive
tests; node --check the plugins).
* fix(grok): deliver intake draft via the relay + correct opencode-mechanism docs
Live end-to-end verification (opencode 1.17.8 + grok-build-0.1) of the WHOLE
integration, then fixes for what it surfaced:
1) Intake draft card (FUNCTIONAL): opencode's synchronous serve reply
(POST /session/:id/message) returns only [step-start, text, step-finish] — it
does NOT include tool-call parts, so the driver could never extract the
propose_draft draft. intake-tools.js now POSTs the draft straight to the
prompter-live relay (/api/prompter/live/{session}/events, the same endpoint
the driver's relay sink uses), so the panel renders the card regardless.
Verified live: grok calls propose_draft -> the relay receives the draft.
2) Correct misattributed opencode "bugs" (DOCS): earlier comments asserted as
general opencode behavior that a provider.xai block / npm override / config
plugin:-array "break" registration. Re-testing showed those were artifacts of
a PROJECT-level .opencode/opencode.json; from the GLOBAL config (which
opencode_config writes) the built-in provider, model resolution, the plugin
array AND the auto-discovery dir all work, and MCP gateway verbs register
(delivery agents verified). Reframed the comments as design choices (built-in
provider + XAI_API_KEY env + plugins baked in the auto-discovery dir with
named exports) and dropped the false claims.
3) Reasoning --variant: passing it does not error, but whether opencode applies a
named reasoning variant to grok-build-0.1 (no provider-defined variants) is
UNVERIFIED — comment softened from a "~54% cut" claim to best-effort,
measure-on-NAS.
Verified live this session: one-shot delivery (model + MCP verbs + plugins +
hooks), secretary tools (read_company_state + submit_directive -> backend with
token), intake draft (relay), grok built-in-provider tool-calling. Remaining
NAS-only: full container assembly (SDK :9000 startup, entrypoint hooks, 429
parking) + the --variant cost measurement. Gate green (ruff/mypy + 51 tests;
node --check the plugins).
* feat(grok): reap abandoned interactive chats (M3)
An interactive intake/secretary chat the human abandoned (closed the tab without
confirming or stopping) leaked its container until the orchestrator restarted —
the wedged-grok reaper is task-driven and these run task_id=None, and an SSE
disconnect intentionally does NOT reap (so a page reload can reconnect).
Reap by IDLE TIME, not connection state: PrompterLiveRegistry tracks
last_activity (bumped on every push/deliver = a turn), and the 60s sweeper
retires sessions idle past ROBOCO_INTERACTIVE_IDLE_REAP_SECONDS (default 1800;
0 disables) via reap_intake_session / reap_secretary_session. An active or
page-reloaded chat that keeps exchanging turns stays fresh and is never reaped;
board-review-parked sessions (task_id set) are exempt. Provider-agnostic — fixes
the leak for both Claude and Grok interactive.
Tests: idle-only reap (active/parked/closed excluded), activity bump keeps a
session alive, threshold 0 disables. Gate green (ruff/mypy/xenon + prompter_live).
* fix(panel): resolve agent names from the live roster so they never drift
A review task assigned to the pr-reviewer rendered as a truncated raw
UUID instead of its name. Root cause: the panel resolved assignees from a
hardcoded static roster in agent-utils.ts that had drifted — it never
gained the board-adjacent agents added backend-side (intake-1,
secretary-1, pr-reviewer-1). Their UUIDs hit no map entry, so
getAgentDisplayName fell through to the unknown-UUID branch and returned
agentId.slice(0, 8). Every assignee surface (task table, task detail,
subtasks, journals, communications, commit cards) shares that resolver, so
all of them showed the fragment.
Make the live /api/agents roster the source of truth instead of a static
duplicate that silently rots:
- agent-utils: add a runtime registry (registerAgentRoster) keyed by both
UUID and slug; resolveToSlug / getAgentDisplayName / isKnownAgent consult
it first. The static maps remain only as an offline / first-paint
fallback (now complete with the three agents).
- api/agents: surface the backend UUID on AgentDefinition (getAll/getOne
previously dropped it), so the registry can key by UUID.
- use-agents: add useAgentRosterSync (registers the live roster) and derive
useAgents from live definitions, falling back to the static roster.
- providers: mount the sync once inside QueryClientProvider.
Now any agent the backend knows about resolves, including ones added after
this change — the panel can no longer drift out of sync.
Tests: agent-utils unit tests cover the three agents end-to-end, a
live-roster-only agent (drift-proofing), live-overrides-static, and a
regression guard for the existing roster.
* fix(pr-review): post a COMMENT review when GitHub forbids self-review
A pr-reviewer review of an org-authored PR never reached GitHub. The agent
side ran correctly (claim → read-only diff → review → post_pr_review →
completed + CEO notify), but the GitHub publish 422'd with "Can not request
changes on your own pull request": the PR was authored by the same account
that owns the project PAT. post_pr_review posts best-effort after the DB
transition, so the failure was logged and swallowed — the task completed and
the CEO was notified "reviewed" while the PR showed no review.
GitHub forbids APPROVE / REQUEST_CHANGES on your own PR but DOES allow a
plain COMMENT review. The org's internal PRs (and any PR the PAT owner
opened) hit this. Retry once as a COMMENT review on the self-review 422 so
the review actually lands; the verdict is already stated in the body. The
external/fork-PR path (different author) is unchanged — REQUEST_CHANGES
succeeds there and the fallback never fires.
Tests: self-review 422 downgrades to COMMENT and returns the COMMENT result;
a failing COMMENT retry still surfaces GitError with no infinite loop; the
existing non-self 422 still raises.
* fix(grok): harden cost-guard, pin runtime, refresh stale plugin comments
Address review findings on the Grok provider work:
- budget-feed plugin failed open unconditionally, so a one-shot task agent
whose in-container SDK budget server went unreachable would run with the
cost cap unenforced. The entrypoint now exports ROBOCO_BUDGET_ENFORCE=1
(one-shot agents always start that server) and the plugin's pre-exec gate
fails CLOSED when the flag is set and the budget endpoint is unreachable,
halting an uncapped burn. Interactive serve agents (intake/secretary) set
no flag and keep failing open (they run no budget server by design).
- Pin opencode-ai to the live-verified 1.17.8 (was an unpinned global npm
install). Untrusted model output runs under it; bump the pin deliberately.
- Document the ROBOCO_GROK_* operator vars in .env.example (image, the three
opencode permissions, reasoning effort, idle-kill, cost ceiling).
- Refresh stale plugin comments: the MCP tool-name shape and the secretary
tool-registration path are confirmed live, and secret-scrub's load route is
the auto-discovery dir (not a config plugin: array). Keep the honest
not-yet-exercised caveat on secret-scrub's deny path and the reasoning
variant — those remain genuinely unverified.
* fix(grok): unbreak workspace-cwd agents, free trapped agents, stop self-PR review
Three bugs surfaced by the first live Grok lifecycle run:
- Dev/QA/doc agents crash-looped at startup with ModuleNotFoundError on
roboco.llm.providers. The entrypoint ran the opencode-config render from the
agent's workspace-clone cwd, whose own roboco/ dir shadows /app on the
sys.path front; a branch without the grok code lacks the providers package.
Render from /app so the installed package always resolves (the render has no
cwd dependency — writes global, reads ROBOCO_MCP_CONFIG).
- A budget/loop halt blocked EVERY tool, including i_am_idle, unclaim, and
i_am_blocked, so a halted agent could neither continue nor stop and flailed —
one billed model turn per blocked retry. The before-gate now always lets the
release verbs through so a halted agent can exit cleanly.
- The inbound reviewer ingested the org's OWN PRs (authored by the repo-owner
account), which can't take a REQUEST_CHANGES review (GitHub 422) and get
re-reviewed every poll. The normalizer flags author_is_owner and ingestion
skips them — the reviewer reviews only PRs the org did not author.
External/contributor PRs are unaffected.
Tests: owner-authored PR flagged + skipped; normalize shape covers the new
field. Gate green on the changed modules (ruff/mypy/xenon + 48 tests).
* feat(grok-cli): render config.toml + map per-role grok CLI flags
First piece of the Grok CLI provider that replaces the opencode runtime: a
pure, unit-tested module the agent entrypoint runs to translate the mounted
mcp-config.json into ~/.grok/config.toml ([mcp_servers]) and compute the
per-role 'grok -p' flags — subagent/shell/edit tool removal, raw-git-mutation
and rm-rf denies, reasoning effort — mirroring ClaudeCodeProvider's per-role
permissions with native grok flags instead of an opencode permission block +
JS guard plugins. Uses tomli_w. The rendered config + env injection are
validated live against grok-build (the model called the server through it).
* feat(grok-cli): grok CLI agent image + headless entrypoint
The roboco-agent-grok image now installs xAI's official grok CLI (Grok Build,
pinned 0.2.56) instead of opencode, authenticated by the SuperGrok subscription
via a mounted ~/.grok/auth.json (parity with the Claude ~/.claude mount, no
metered API key). The entrypoint renders ~/.grok/config.toml + per-role flags
from /app (the ModuleNotFound-shadowing lesson), runs grok -p headless with
--output-format json, keeps the prompt-injection guard, and exits 75 on a
rate-limit so the orchestrator parks the provider. No in-container SDK server or
budget-feed — native --max-turns + server-side terminal-substitute replace them.
* feat(grok-cli): GrokCliProvider — subscription auth mount, mirrors ClaudeCodeProvider
Replace the opencode GrokProvider with GrokCliProvider: reuses the orchestrator's
shared mount/auth/git assembly (gateway + identity) exactly like the Claude path,
mounts the host ~/.grok/auth.json read-only (SuperGrok subscription) instead of
injecting an xAI key, and sets the slim env the grok-cli entrypoint + renderer
read (ROBOCO_AGENT_ID for per-role flags, model, mcp-config, prompt). Provider
routing fields are blanked before the shared step so the grok endpoint is never
mislabelled ANTHROPIC_*. Per-role permission logic now lives in grok_cli_config,
so the provider is slim. Registry/orchestrator/exports updated; provider tests
rewritten for the CLI behavior (no XAI key, auth mount present/absent).
* feat(grok-cli): capture per-session token usage + notional cost
Grok runs on the SuperGrok subscription, but — exactly like Claude on Max — we
still record per-agent tokens and a notional cost for the dashboard. The grok
CLI writes a cumulative totalTokens per turn into
~/.grok/sessions/<cwd>/<session-id>/updates.jsonl (the grok analogue of the
Claude transcript / old opencode.db); the max is the session total. This reader
locates that file (url-encoded cwd), extracts the total, and prices it at the
output rate (no input/output split from the CLI; conservative + matches the
reasoning-at-output convention). Validated against a real grok-build session
(18253 tokens -> $0.0365). Entrypoint + finalize wiring follows.
* feat(grok-cli): wire usage capture into the run (session id + post-run extract)
The provider pins a fixed session id (ROBOCO_AGENT_SESSION_ID, reused from the
agent session id as on the Claude path); the entrypoint passes it to
'grok -p -s <id>' so the run's session store is locatable, then runs the usage
reader post-run (best-effort) to write the captured tokens + cost. The
orchestrator-side finalize that reads that file follows.
* feat(grok-cli): read captured usage at finalize; keep interactive serve working
The provider mounts the per-agent data dir and points the entrypoint's usage
file at it; the orchestrator's grok finalize reads that usage.json first (the
grok-CLI total, priced at the output rate) and falls back to opencode.db for the
still-opencode interactive intake/secretary path. Re-add _reasoning_effort_for to
grok.py as a clearly-temporary shim for that interactive path (it needs opencode's
"minimal" variant, distinct from the CLI's --effort) until it is converted too.
* feat(grok): convert interactive intake/secretary to the grok CLI; delete opencode
Move the last Grok runtime off opencode onto xAI's official `grok` CLI, for full
parity with the Claude path. The intake/secretary chat now runs per-turn headless
`grok -p` invocations that resume one session id (proven live: context carries
across runs), with streaming-json deltas mapped to the existing panel StreamChunk
kinds — the IntakeDriver loop, message source, relay, and idle reaper are reused
unchanged; only the SessionFactory differs (GrokCliSession replaces the
opencode-serve session).
- GrokCliSession + a pure, unit-tested streaming-json -> StreamChunk assembler
(thought coalesced to one block, text streamed live, end captures the session
id for -r, fenced-draft fallback, clear errors incl. rate-limit).
- intake propose_draft and secretary read_company_state/read_task/submit_directive
are now FastMCP servers (roboco-intake / roboco-secretary) wired into
~/.grok/config.toml, launched via `uv run --directory /app` to resolve the
installed package. The secretary tools reuse the shared backend helpers.
- Orchestrator: interactive spawn mounts the subscription auth + per-agent usage
dir (no metered xAI key, no permission env — grok flags carry per-role perms);
usage/cost now read a captured usage.json (drop the opencode.db reader, the
_opencode_db_path/_grok_usage_from_opencode methods, and the cost-cap's
opencode read). hosts["opencode"] -> hosts["grok_usage"]; OPENCODE_DATA_DIR ->
GROK_USAGE_DATA_DIR.
- Fix one-shot usage capture: `-s` does not pin the session id (grok generates
its own), so the entrypoint now reads the real id back from the JSON run log
and the reader uses it; usage is captured per-turn on the interactive path.
- Delete the opencode layer: opencode_config/opencode_usage/opencode_session, the
docker/grok/*.js plugins, the old one-shot entrypoint, and their tests.
- Compose (all three files), .env.example, and stale comments updated to the
grok-CLI runtime; add the SuperGrok auth mount + grok-usage dir.
Gate green: ruff, mypy (296 files), xenon, tests. NAS build/verify pending.
* fix(grok): deliver the role blueprint as grok's system prompt via ~/.grok/AGENTS.md
The blueprint was mounted at /app/system-prompt.md but never reached grok — a real
parity gap vs the Claude path (which passes --system-prompt-file). grok agents ran
only on the per-task prompt, missing their RoboCo role/org context.
Verified live on grok 0.2.56 that the obvious flags do NOT work headless:
`--system-prompt-override` and `--rules` are silently ignored under `grok -p`
(identical output with and without). What IS honoured is grok's instruction-file
discovery — and `$HOME/.grok/AGENTS.md` is loaded GLOBALLY regardless of --cwd
(a project AGENTS.md only loads from the cwd/project root, which would pollute the
agent's git workspace). Proven end to end: a blueprint written there makes grok
adopt the role ("I am the RoboCo intake interviewer ... -- intake-1").
write_agents_md() copies /app/system-prompt.md -> ~/.grok/AGENTS.md; the one-shot
render (grok_cli_config.main) and both interactive mains call it. No git pollution
(it lives in ~/.grok, not the workspace), and it covers repo-cwd and /app-cwd
roles alike. Reverted the non-working --system-prompt-override wiring.
* feat(grok): close the Claude-parity divergences (reasoning, subagents, web, bash-guard)
Bring the grok CLI to parity with the Claude path on the four deliberate
differences:
- Reasoning: drop the per-role `--effort low` default — Claude sets no per-role
thinking budget, so grok now uses the model default for every role. The
fleet-wide ROBOCO_GROK_REASONING_EFFORT override stays as a cost lever. (This
also un-caps intake-draft quality, the one that actually mattered.)
- Subagents: the intake interviewer may now fan out to subagents (parity with the
Claude intake's `Task` allowance); every other role still has `Agent` removed.
- Web: `--disable-web-search` for every role — no agent gets direct web (Claude's
tool set has none either); the roles that get web reach it through the gated
roboco-search MCP, unaffected.
- Bash command filtering: full parity, split by deny semantics. Verified live that
a grok PreToolUse hook deny CANCELS the run, while native `--deny` denies
GRACEFULLY (the agent gets a permission error and recovers). So:
* git network/branch/history ops -> native `--deny` (operational reflex; the
agent must recover, not drop the task). Expanded to the full bash-guard set.
* credential-exfil / identity-forgery / internal-API / env-dump patterns ->
the SAME bash-guard the Claude path runs, wired as a grok PreToolUse hook
(ROBOCO_GUARD_SKIP_GIT=1 so it leaves git to `--deny`). A hard cancel is the
right response there — no legitimate agent reads ~/.netrc or forges an
X-Agent-ID. One tolerance line (accept grok's camelCase `toolInput`) makes
the one tested script guard both runtimes; +5 grok cases (50/50 green).
Also cleaned stale internal task-number / smoke labels out of bash-guard-hook.sh.
* fix(grok): install grok CLI to ~/.grok/bin (its real default), not ~/.local/bin
The image build failed at `chown ... /home/agent/.local: No such file or
directory`. The grok installer's default is $HOME/.grok/bin — the binary lands at
~/.grok/bin/grok; ~/.local/bin/grok is only a convenience SYMLINK the installer
creates on macOS but not in the Linux container. So the Dockerfile referenced a
directory that never existed:
- PATH pointed at ~/.local/bin -> `grok` would not be found at runtime even if
the build had passed;
- chown targeted ~/.local -> the build aborted.
Point PATH + chown at ~/.grok/bin / ~/.grok. Also harden the install: download the
script to a file (a `curl | bash` pipe swallows a curl failure as a silent no-op)
and verify the binary installed and runs (`test -x` + `grok --version`), so a
broken install fails the build loudly instead of producing a grok-less image.
* fix(grok): address adversarial-review findings across the grok-CLI conversion
A 7-dimension adversarial review (find -> independently refute) surfaced 14 real
issues; fixed each:
Runtime bugs
- GrokCliSession.send drained stdout fully BEFORE stderr — a >64KB stderr burst
would deadlock the turn forever (spinner never clears). Drain stderr
concurrently, and add a per-turn watchdog (ROBOCO_GROK_TURN_TIMEOUT_SECONDS,
default 600s) that kills a wedged process and emits error+turn_end.
- Crash-restarted grok agents launched `grok -p ""` (empty prompt) — Claude gets
a scan-for-work fallback. Default the prompt in _spawn_container so every
dedicated provider gets it too.
- _grok_usage_json read /data/grok-usage unconditionally while its writers branch
compose-vs-local, so a local-mode agent finalized at $0 and the cost-cap was
inert. Single-source the path in a new _grok_usage_dir helper (read == write).
- GrokCliSession secretary role fell through to "unknown" (get_agent_role returns
a truthy sentinel, never None), defeating the ROBOCO_AGENT_ROLE fallback.
Parity / hardening
- --deny set was missing `git tag -d` / `git reflog delete` that the Claude
bash-guard blocks — added them (the "same set" claim is now true).
- Interactive mains now install the bash-guard hook too (defense-in-depth).
- Compose: collapse the GROK_AUTH_DIR / ROBOCO_HOST_GROK_DIR auth-mount pair into
one canonical var so a partial override can't silently break agent auth.
Docs / comments
- Panel routing card + architecture security doc no longer say Grok runs on the
deleted opencode runtime; orchestrator comments point at the renamed entrypoint.
Tests
- Cover the interactive _render_grok_config MCP wiring (ModuleNotFound guard +
secretary HMAC env), the cost-cap kill-failure + interactive relay-close paths,
the local-mode usage read, the role fallback, the turn timeout, and the new
git denies. (#13 — a separate grok "Write" tool — investigated: grok's only
built-in file-mutation tool is search_replace, already removed; no gap.)
Gate green: ruff, mypy, xenon, tests.
* fix(grok): declare tomli-w as a runtime dependency (agent image needs it)
The grok agent image failed at spawn with `ModuleNotFoundError: No module named
'tomli_w'` when rendering ~/.grok/config.toml. tomli_w was only a transitive dep
of a dev-extra package, so it was present in dev/orchestrator envs but excluded
from the agent image, which builds its venv with `uv sync --frozen --no-dev`.
grok_cli_config imports it at module load to serialize the MCP gateway config, so
without it a Grok agent gets no gateway verbs.
Promote tomli-w to a direct [project.dependencies] entry. Locked with
`--upgrade-package tomli-w` so only tomli-w is added — no incidental churn of the
8 unrelated packages a full re-resolve would have bumped.
* Updated uv.lock
* fix(grok): auto-approve tool execution (--always-approve) so headless agents can call tools
Live smoke caught every grok agent (Main PM, pr-reviewer, dev, …) ending its run
with stopReason=Cancelled and empty output the instant it reached for a tool. Root
cause: headless `grok -p` cannot approve a tool call without `--always-approve`
(grok's docs: required for unattended automation), and the per-role args didn't
pass it — so no agent could call a gateway verb, an edit, or an MCP tool, and the
run was cancelled.
Add `--always-approve` to grok_cli_args_for_role (one place → every role, one-shot
and interactive). Safety is unaffected: `--disallowed-tools` still removes tools
and `--deny` still hard-blocks command patterns regardless of approval (a denied
command returns a permission error and the agent recovers — verified live).
Proven in the rebuilt image side-by-side: without the flag a tool call yields
Cancelled/not-called; with the real rendered args it returns EndTurn and the MCP
tool actually runs. (My earlier in-image tool-calling check passed `--always-approve`
manually, which masked that the production args omitted it — fixed.)
* fix(pr-review): seed claim heartbeat so the grok reviewer isn't wedge-killed
pr_review_claim transitioned a review task pending -> in_progress but never
seeded last_heartbeat_at, unlike every sibling claim path (_finalize_claim,
qa_claim, doc claim). The reaper treats a NULL heartbeat as a stale claim, and
the GROK idle-kill watchdog bypasses the live-container skip on a NULL
heartbeat -- so the reviewer container was killed (Cancelled) before it could
post_pr_review, churning the task back to pending on a respawn loop. A Claude
reviewer was shielded by the live-instance skip; only GROK manifested it.
Seed the heartbeat at claim time, matching the established invariant. Verified
against a real Postgres (10/10 test_pr_review_db tests, incl. the new
last_heartbeat_at assertion).
* fix(grok): stream one-shot output live + capture real token usage
Two gaps the buffered run hid, both verified in the real image with mounted
SuperGrok auth:
- Observability: the entrypoint buffered grok's output to a temp file and only
cat it after the run, so `docker logs` was blank while the agent worked.
Switch the one-shot to --output-format streaming-json piped through tee:
grok flushes each thought/text event incrementally (confirmed token-by-token
live in-container), so the agent's reasoning shows in docker logs in real
time, parity with the Claude stream-json path. Read the session id back from
the NDJSON run log (the terminal `end` event) since -s does not pin it.
- Usage: total_tokens read 0 for every grok run. grok nests the cumulative
totalTokens on params.update._meta, but the reader looked at params._meta
(which only holds event ids); the unit fixture had the same wrong shape, so
the tests masked it. Read the real path (with params._meta / top-level
fallbacks) and fix the fixture to the real grok shape. Verified live:
usage.json now reports total_tokens=3262, cost_usd=0.006524 (was 0).
* fix(grok): validate agent_id before using it as a usage-dir path segment
CodeQL flagged a high-severity py/path-injection: agent_id flowed from
request-facing call sites into _grok_usage_dir() and on to read_text(), so a
value containing '..' or a separator could traverse the filesystem. Validate
agent_id against the slug/uuid allowlist ([A-Za-z0-9_-]+) at the single
chokepoint (_grok_usage_dir feeds both the mount and the finalize read);
anything else raises. Rejects traversal; accepts every real agent slug.
* fix(grok): use explicit-guard path sanitizer CodeQL recognizes as a barrier
The re.fullmatch allowlist from
|
||
|
|
ff35a646fa |
Chore: reduce analytics complexity (#100)
* refactor(analytics): reduce cyclomatic complexity in usage/pricing/rollup
Collapse the three near-identical get_by_* aggregation methods in
UsageService into a shared _aggregate_by helper parameterized by group
column and key name, and centralize token null-coalescing in a
_row_tokens helper. Extract the per-row upsert in _sweep_daily_rollup
into _upsert_rollup_row, and the pricing-table lookup into
_lookup_prices. All blocks now rank <= B and both modules rank A, so the
xenon gate passes; behavior is unchanged and existing tests stay green.
* feat(billing): make token pricing provider-aware
Distinguish three cases when a model has no per-token rate: a non-Anthropic
model (local Ollama, or an Ollama Cloud ":cloud" model billed by flat
subscription / GPU-time) legitimately has no per-token cost and returns 0.0
silently; an unpriced Anthropic ("claude"-named) model also returns 0.0 but
logs a warning, since that is real spend being undercounted and catches new or
renamed Claude models missing from the table. Folds the old ollama/ prefix
special-case into the general non-Anthropic path so there is one code path,
and replaces the blanket 'no pricing data' warning that fired even for
self-hosted models.
* fix(tasks): preserve ownership when force-unclaiming to pending
The stale-claim reaper and the dependency-blocked release both routed through
_force_unclaim_to_pending, which nulled assigned_to and left the task in a
pending state owned by nobody — no dispatcher re-spawns an ownerless pending
task, so it went dormant. The dispatcher-side claimed_by fallback only masked
half the cases.
Capture the owner before releasing the claim and keep both assigned_to and
claimed_by pointed at it (mirroring the unblock restore), releasing only the
live claim (active_claimant_id + heartbeat) and the WorkSession. The same agent
now resumes the task once it re-dispatches. Updates the reaper test that
asserted the old orphaning behavior and adds owner-preservation coverage for
both the reaper and dependency-release paths.
* fix(tasks): unblock restores the owner into both ownership fields
Audit follow-up to the force-unclaim ownership fix. unblock() only restored
assigned_to from blocker_raised_by, which block() stashes solely from
assigned_to. A task claimed via give_me_work (claimed_by set, assigned_to null)
therefore unblocked into a split-owner state — assigned_to null but claimed_by
set — that both the dev dispatcher and the PM pool-router race to pick up. It
also left claimed_by pointing at the resolver PM after an escalation.
Resolve the owner as blocker_raised_by or assigned_to or claimed_by and write
it to both fields, matching the force-unclaim and reassign convention so the
original worker resumes cleanly. Adds coverage for the give_me_work-claim case
and asserts owner restoration on the existing in_progress-resume test.
* test(orchestrator): cover dev owner resolution and the claimed_by fallback
_resolve_dev_owner_uuid had no coverage. Add the status-dependent precedence
(claimed/blocked prefer the live claimant; other statuses prefer the
PM-assigned owner) and the half-reap fallback where a pending task with
assigned_to nulled still resolves its owner from claimed_by instead of going
dormant.
* fix(tasks): wire the pre-block snapshot so unblock(restore=True) works
The restore=True path on a PM unblock was a no-op: pre_block_state /
pre_block_assignee (migration 006) were read by unblock_with_restore but never
written, so it always fell through to legacy unblock() and the restore flag did
nothing.
Snapshot the resting status + owner at every block entry (dependency block,
soft block, escalation) before mutating, capturing only the first block in a
chain so a re-block doesn't overwrite the original state. Escalation snapshots
the outgoing owner, not the escalation target, so restore returns the original
worker. The restore path applies the same branchless guard legacy unblock()
relies on — a snapshotted in_progress with no branch diverts to pending instead
of looping the dispatcher — and is extracted into _apply_pre_block_restore to
keep complexity under the gate. Adds coverage for snapshot capture, restore,
the branchless divert, and escalation owner restoration.
* test(tasks): update orphan-reconciler and dependency-release tests for owner preservation
Both the startup orphan reconciler and the dependency-blocked claim release
route through unclaim_for_reaper / _force_unclaim_to_pending, which now preserve
the owner instead of nulling assigned_to. Update the two tests that asserted the
old orphaning behavior to assert the owner is kept (so the same agent resumes)
while the live claim is released.
* chore(tests): scrub internal work-item labels from test names, docstrings, comments
Rename four test files that carried audit work-item IDs in their filenames
(test_p0_7_branch_atomicity, test_p2_8_orphan_reconciler,
test_p2_9_autogen_prompt_layer, test_p2_7_attempt_id) to describe what they
test, and strip the matching P-/D-/S- cluster labels from docstrings, comments,
and assertion messages across the test suite and two orchestrator comments.
These are internal references with no meaning in the codebase; behavior is
unchanged.
* style: reformat assertion line shortened by the internal-ref scrub
* build: waive unreachable torch CVE-2025-3000 in pip-audit gate
torch is a transitive CPU-pinned dep (piragi / sentence-transformers) never
loaded at runtime — the stack uses Ollama over HTTP for all embeddings/LLM, so
the vulnerable torch.jit.script path is unreachable. CVE-2025-3000 is MEDIUM,
local-only, with no published fix. Documented --ignore-vuln waiver; revisit when
a fixed torch ships.
* fix(orchestrator): route unplaceable pending tasks to main-pm instead of dropping them
_get_routing_target returned None when a 'dev'-classified task had no cell
agent (no team, or a non-cell team like fullstack/system) or when the routing
classification was unrecognized. _route_unassigned_pm_task logged 'no routing
target found' and returned, leaving the task ownerless and pending — and no
dispatcher re-spawns an unrouted pending task, so it went dormant for 10+ min
until the stuck-task detector caught it.
Fall back to main-pm (the same default cell_pm routing and escalation already
use) so the task is always owned and triaged, never stranded. Logs the fallback
so unplaceable tasks stay visible. Adds a test asserting no (routing, team)
combination ever resolves to None.
* fix(panel): make intake chat markdown inherit the bubble's text color
MarkdownBody is shared by the assistant (text-foreground) and user
(text-primary-foreground) bubbles. [&_*]:!text-inherit only colored the prose
div's descendants, so the prose div itself kept the prose typography body color
(gray) and children inherited that — unreadable on the muted assistant bubble.
Add !text-inherit on the prose div itself so it inherits the bubble's color
too; descendants then inherit the correct foreground. Fixes both bubbles without
hardcoding a color.
* fix(prompter): keep a board-reviewed product on the board team so Approve & Start shows
A product coordination root confirmed via 'Board review & Start' is assigned to
a board reviewer (product-owner) for review, but create_task_from_draft set
team=main_pm for every product unconditionally. The CEO's Approve & Start gate
keys on team=board, so the button never appeared — and because the owner stayed
a board agent while the team said main_pm, the dispatcher routed it to the board
path (nothing left to do after review) and the task stranded at pending, with
the board agent fruitlessly trying to escalate it up.
Route a product by its assignee: a board reviewer keeps it team=board (so the
gate appears and approve_and_start later hands it to Main PM), while a main-pm
assignee — the 'Approve & Start' straight-through path — is team=main_pm. Adds
_assignee_is_board mirroring TaskService's board-role check, and a test.
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
|
b3057628b0 |
[499f9eb1] Token Usage & Cost Analytics — Full-Stack Instrumentation, Persistence, and Visualization (#90)
* [cd2bf666] feat(usage): add token usage types, API client, hooks, and UI components (#87) (#88) - Append 5 TypeScript interfaces to src/types/index.ts: TokenUsageSnapshot, AgentUsageRow, UsageSession, UsageTimePoint, ModelUsageSlice - Create src/lib/api/usage.ts: Axios singleton + isMockMode guards for getUsageSnapshot, getUsageTimeSeries, getAgentUsage, getUsageSessions, getModelUsage - Create src/hooks/use-usage.ts: usageKeys factory + useUsageSnapshot, useUsageTimeSeries, useAgentUsage, useUsageSessions, useModelUsage hooks - Create UsageOverviewPanel (dashboard/usage-overview-panel.tsx): 6 metric rows with Skeleton loading state; week-over-week trend arrow for cost - Update CommandCenter: Metrics+Alerts row expanded from 2-col to 3-col grid adding UsageOverviewPanel - Create src/components/metrics/ folder: UsageTimeSeriesChart (recharts stacked AreaChart with var(--chart-1/2/3)), ModelUsageDonut (PieChart), AgentUsageChart and TeamUsageChart (BarChart), SessionsTable (sortable columns + 10-row Prev/Next pagination) - Update Metrics page: Token Usage & Costs section with 5 rows (summary cards, time series+donut, agent+team bar charts, projection+cache efficiency, sessions table) - Add usage mini-bar to AgentCard: token count + cost + progress bar; AgentGrid and Agents page pass agentUsageMap through - Install recharts 3.8.1 - Export all new symbols through their barrel index.ts files Co-authored-by: Frontend Developer 1 <fe-dev-1@agents.roboco.dev> * [10372f0f] Implement full token usage instrumentation: DB migration, SDK endpoints, orchestrator hooks, analytics API, WebSocket events, dashboard integration (#86) (#89) * [10372f0f] feat(token-usage): add Alembic migration 026 for token usage tables Create agent_spawn_sessions, token_usage_snapshots, and daily_usage_rollups tables with correct BIGINT columns, indexes, and unique constraint. Chain: 025_agentrole_prompter → 026_token_usage_tables. * [10372f0f] feat(token-usage): add ORM table classes for token usage instrumentation Add AgentSpawnSessionTable, TokenUsageSnapshotTable, DailyUsageRollupTable to db/tables.py. Import BigInteger and Date from SQLAlchemy. All columns match the migration schema with BIGINT token counts and proper indexes. * [10372f0f] feat(billing): add pricing module with calculate_cost() function Create roboco/billing/__init__.py and roboco/billing/pricing.py with calculate_cost() supporting Claude opus/sonnet/haiku models with input/output/cache pricing. Unknown models return 0.0 without raising. * [10372f0f] feat(sdk): add POST /usage/report and GET /usage/status endpoints to agent SDK Extend _SessionState with token counters. Add TokenReportRequest and TokenUsageStatus models. POST /usage/report additively accumulates token counts; GET /usage/status returns current session totals for sweeper polling. * [10372f0f] feat(orchestrator): add token usage instrumentation hooks - _launch_spawn() calls _record_spawn_session() after successful container spawn - stop_agent() calls _finalize_spawn_session() before container removal - _run_sweep() calls _sweep_token_snapshots() and _sweep_daily_rollup() each tick - New methods: _record_spawn_session, _finalize_spawn_session, _sweep_token_snapshots, _sweep_daily_rollup in TOKEN USAGE section * [10372f0f] feat(api): add token usage analytics API with 7 endpoints Create roboco/services/usage.py (UsageService) and roboco/api/routes/usage.py. Endpoints: GET /api/usage/summary, /time-series, /by-agent, /by-team, /by-model, /projection, /cache-efficiency. Register in app.py. * [10372f0f] feat(dashboard): add usage_summary field to CEO dashboard Add UsageSummary schema (tokens_today, cost_today_usd) to dashboard schemas. Add usage_summary: UsageSummary | None to CEOOverview. Update get_ceo_overview() to populate usage_summary from daily_usage_rollups. * [10372f0f] fix(billing/tests): remove dead except block in _sweep_daily_rollup, add unit tests for pricing.py and services/usage.py - Remove unreachable `except Exception as e` block in orchestrator.py _sweep_daily_rollup() (lines 3376-3381) which referenced undefined `agent_id` and was copy-pasted from _sweep_token_snapshots by mistake - Add tests/unit/billing/test_pricing.py: 31 tests covering opus/sonnet/ haiku tiers with all 4 token types, unknown model → 0.0, empty string → 0.0, and substring-match priority (longer fragment wins) - Add tests/unit/services/test_usage.py: 25 tests covering get_summary trend_pct edge cases (prev=0, both=0, prev>0), get_by_agent/team/model pct_of_total summing to 100%, get_projection formula (avg_daily×30), and get_cache_efficiency hit-rate and cost_saved arithmetic - pricing.py: 100% coverage; services/usage.py: 83% coverage (>80% target) * [10372f0f] fix(usage): include cache tokens in time-series total_tokens to fix AC9 consistency violation get_time_series() previously computed total_tokens as tokens_input + tokens_output only. get_summary() includes all 4 token types (input + output + cache_read + cache_write). AC9 requires both endpoints to agree on their totals for the same period. Fix: add tokens_cache_read and tokens_cache_write to the SELECT query in get_time_series() and include them in the total_tokens calculation. Also adds 4 new unit tests in TestGetTimeSeries covering: - total_tokens includes cache_read and cache_write (the AC9 guard) - zero cache tokens still produces correct total - empty result returns empty list - required fields are present in each point * [10372f0f] fix(usage): remove unused imports and include cache tokens in breakdown totals (AC10) - Remove import math (F401 — never used) - Remove text from sqlalchemy import (F401 — never used) - Remove unused local calculate_cost import inside get_cache_efficiency (F401) - Add tokens_cache_read and tokens_cache_write to SELECT in get_by_agent, get_by_team, and get_by_model; update grand_total and per-item total to include all 4 token types so totals match get_summary() (AC10 fix) - Update test mock rows to include explicit tokens_cache_read=0 and tokens_cache_write=0 so they work with the fixed code - Add new test cases: test_cache_tokens_included_in_total_tokens and test_pct_of_total_sums_to_100_with_cache_tokens for each breakdown class --------- Co-authored-by: Backend Developer 1 <be-dev-1@agents.roboco.dev> * [44b9eb1f] feat(usage): align frontend API client, TS types, and chart components to real backend contract (#92) (#94) Update all usage-related frontend code to match the actual FastAPI backend response shapes and endpoint paths: - panel/src/lib/api/usage.ts: rewrite all 7 API functions to use correct endpoint paths (/usage/summary, /usage/by-agent, /usage/by-model, /usage/by-team, /usage/time-series, /usage/projection, /usage/cache-efficiency); send period query param (24h/7d/30d not hours); mock generators produce data matching real backend shapes exactly; getUsageSessions returns [] in prod (no /usage/sessions endpoint exists) - panel/src/types/index.ts: replace TokenUsageSnapshot with UsageSummary (tokens_input/tokens_output/total_cost_usd/trend_pct); update AgentUsageRow to use agent_slug/total_tokens/cost_usd/pct_of_total; add TeamUsageRow, UsageProjection, CacheEfficiencyResponse; update UsageTimePoint to use bucket field; update UsageSession to use agent_slug - panel/src/hooks/use-usage.ts: rewrite all hooks to match new API and types; add useTeamUsage, useUsageProjection, useCacheEfficiency hooks - panel/src/components/metrics/usage-time-series-chart.tsx: use bucket field (not timestamp) for axis labels - panel/src/components/metrics/agent-usage-chart.tsx: use agent_slug and total_tokens (not agent_name/tokens_today) - panel/src/components/metrics/team-usage-chart.tsx: rewrite to accept TeamUsageRow[] from API directly - panel/src/components/metrics/model-usage-donut.tsx: use total_tokens, cost_usd, pct_of_total (not tokens/cost/percentage) - panel/src/components/metrics/sessions-table.tsx: use agent_slug, sort keys updated - panel/src/components/dashboard/usage-overview-panel.tsx: use useUsageSummary with tokens_input/tokens_output/total_cost_usd/trend_pct - panel/src/app/(dashboard)/metrics/page.tsx: wire all new hooks, add TeamUsageChart, ProjectionCard, CacheEfficiencyCard with correct types - panel/src/app/(dashboard)/agents/page.tsx: key agentUsageMap by agent_slug - panel/src/components/agents/agent-card.tsx: use total_tokens and cost_usd Co-authored-by: Frontend Developer 1 <fe-dev-1@agents.roboco.dev> * [2161b832] fix: SDK_PORT constant, stop_agent lock refactor, usage_session_id binding, rollup 7-day window (#93) (#95) - Add SDK_PORT = 9000 module-level constant to orchestrator.py; replace hardcoded 9000 in _sweep_budget_exceeded URL with SDK_PORT - Add UUID to TYPE_CHECKING imports to satisfy ruff F821 - Refactor stop_agent: call _finalize_spawn_session BEFORE acquiring self._lock so the SDK HTTP round-trip does not hold the lock - Add usage_session_id: UUID | None field to AgentInstance dataclass - Change _record_spawn_session to return UUID | None; wire return value back to instance.usage_session_id in _launch_spawn - Update _finalize_spawn_session to use WHERE id=usage_session_id for direct session row lookup when usage_session_id is not None - Add started_at >= (now_utc - 7 days) filter to _sweep_daily_rollup aggregate query to avoid re-aggregating all-time history each sweep Co-authored-by: Backend Developer 1 <be-dev-1@agents.roboco.dev> * [2e0759e1] fix: pricing accuracy, import ordering, session-id binding, rollup cleanup, write-hook tests (#97) (#98) - pricing.py: correct claude-opus-4 prices (5/25/0.50/6.25 not 15/75/1.5/3.75) and haiku family prices (1/5/0.10/1.25 not 0.8/4/0.08/0.20); add Ollama zero-cost early-return; add structlog warning for unmatched model names - app.py: move usage_router import before routes.v1 block (ruff isort fix) - orchestrator.py _sweep_daily_rollup: remove unused calculate_cost import; add blank line between stdlib (uuid4) and third-party (sqlalchemy) imports - orchestrator.py _sweep_token_snapshots: prefer direct lookup by instance.usage_session_id; fall back to agent_slug heuristic only when None - tests: add test_sweep_daily_rollup_inserts_new_row and test_stop_agent_finalizes_before_lock to test_orchestrator_write_hooks.py - usage.py, routes/usage.py, stream_bus.py, test files: ruff format/lint fixes Co-authored-by: Backend Developer 1 <be-dev-1@agents.roboco.dev> * Mypy compliance * fix(migrations,tests): linearize forked migration chain + correct ceo_reject coordination-root expectation The master merge brought in 026_completed_dependency_ids alongside the rework's 026_token_usage_tables — both off 025, forking the alembic head and breaking the enum-parity test. Rebase token-usage onto 026_completed_dependency_ids (linear chain, single head). Also: test_ceo_reject_routes_coordination_task_to_main_pm asserted the old NEEDS_REVISION behavior; the lifecycle fix correctly routes a coordination root to PENDING (Main PM's claim source). Update the assertion. --------- Co-authored-by: Frontend Developer 1 <fe-dev-1@agents.roboco.dev> Co-authored-by: Backend Developer 1 <be-dev-1@agents.roboco.dev> Co-authored-by: Renn F <rennf93@users.noreply.github.com> |