mirror of
https://github.com/rennf93/roboco.git
synced 2026-08-03 07:23:24 +02:00
11c1d9eee79c02802429246f5b89e291ffe9504d
78
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Renn F
|
11c1d9eee7 | feat(content): structured PR-review findings + generated GitHub comment | ||
|
|
5fe1e6df58 |
feat: in-path PR-review gate — per-cell + main reviewers (#229)
* feat(lifecycle): add the in-path PR-review gate status + reviewer verbs
Insert awaiting_pr_review between the assembled-PR submit and the PM merge,
giving the merge level the rejection capability it structurally lacks — today
only qa_fail and ceo_reject ever reach needs_revision, so a PM review is a
merge button with no teeth.
- New Status awaiting_pr_review + submit_for_review / pr_pass / pr_fail actions
(pr_pass -> awaiting_pm_review, pr_fail -> needs_revision, mirroring the QA gate).
- Reviewer verbs claim_gate_review / pr_pass / pr_fail, and a main-PM submit_root
verb (the root analogue of the cell PM's submit_up; opens the root->master PR).
- Extend the self-review-symmetry validator to the new sign-off actions.
- Mirror the value into the ORM TaskStatus enum + the A2A state map, and add the
postgres taskstatus enum value (migration 040, forward-only like 037).
- Regenerate the per-role verb tables; add gate spec tests.
Spec surface only; the gateway methods + dispatch are wired in follow-ups, so the
verbs are advertised but dormant (flow_server tolerates unregistered verbs).
* feat(identity): add the three cell PR-review-gate reviewers
The in-path gate needs a reviewer per cell so each cell's assembled cell->root
PR is reviewed by a stack-specialized agent, while pr-reviewer-1 serves the
root->master gate (and keeps doing inbound external PRs).
- be/fe/ux-pr-reviewer: PR_REVIEWER role, team-scoped (so dispatch routes each
cell's gate to its own reviewer); seeded identities + ROLE_TEAM_RULES + names.
AI agent count 22 -> 25.
- They reuse the existing roboco-agent-pr-reviewer image (AGENT_IMAGES maps the
three slugs to it, as be-dev-1/-2 share one image) — no new image.
- Tracing table: pr_pass/pr_fail require a learning entry (parity with
post_pr_review), submit_root mirrors submit_up, claim_gate_review is waived
(its tracing applies on pr_pass/pr_fail) — completes the verb surface added
in the prior commit.
- Update the roster-pinning identity tests.
* feat(gateway): wire the in-path PR-review gate end to end
Make the assembled-PR review gate operational across the choreographer, the
TaskService transitions, and the v1 flow surface.
- TaskService: submit_for_review (in_progress→awaiting_pr_review), pr_gate_claim
(no-transition reviewer claim), pr_pass (→awaiting_pm_review), pr_fail
(→needs_revision); mirror qa_pass/qa_fail (clear claim, actor-mismatch warn,
issues appended for the PM's revision). VerbRunner gains the matching atomic
handlers + a create_root_pr side effect.
- Repoint submit_up to compose submit_for_review (cell→root PR enters the gate),
and add a main-PM submit_root verb (opens the root→master PR, enters the gate).
- Split main_pm_complete: a code root must pass the gate first (requires
awaiting_pm_review; rejects an in_progress code root toward submit_root and no
longer reopens the PR), while a branchless coordination root still walks
straight through, ungated.
- PRGateMixin (claim_gate_review / pr_pass / pr_fail) composed onto the
Choreographer; flow_server forwarders + v1 routes (pr_reviewer + main_pm) +
request schemas.
- Tests: gate spec + the updated submit_up / main_pm_complete expectations + new
real-DB integration tests driving submit_for_review→pr_gate_claim→pr_pass and
pr_fail through the real enforcement layer.
* feat(orchestrator): dispatch the in-path PR-review gate
Make the gate live in the dispatch loop.
- _dispatch_pr_gate_work: route awaiting_pr_review tasks to reviewers by level —
a cell→root task to its cell reviewer (be/fe/ux-pr-reviewer), the root→master
task to pr-reviewer-1. The reviewer self-claims via claim_gate_review (no
pre-claim, mirroring the external-PR dispatcher); registered in
_dispatch_all_work. _select_agent_for_cell learns the pr_reviewer role.
- _build_pr_gate_prompt: anchors the reviewer to the parent objective + full
acceptance criteria + the FE<->BE contract, then pr_pass / pr_fail.
- _readiness_check_role_for_status: awaiting_pr_review -> pr_reviewer.
- Fail routing: pr_fail reassigns the failed assembled task to its PM
(_revision_pm_for_task: cell PM for a cell team, Main PM for the root), and the
revision dispatcher is generalized from coordination-roots-only to any
PM-owned needs_revision task so the gate-failed task is re-coordinated instead
of deadlocking.
* docs: document the in-path PR-review gate + the cell reviewers (22→25)
Reflect the shipped gate across the canonical + RAG docs.
- CLAUDE.md: agent count 22→25, the cell reviewers in the org chart, an
awaiting_pr_review state + the gate transitions + a gate note in the lifecycle
section, and submit_root / claim_gate_review / pr_pass / pr_fail in the verb
surface table.
- docs/rag/architecture: org-structure (count, cell-reviewer roster, cells
table), agent-uuids (be/fe/ux-pr-reviewer rows), agent-model (role + team
rows).
- docs/rag/roles/pr-reviewer: the in-path gate section + the gate verbs.
- Wrap reviewer.id with UUID(str(...)) in the gate DB tests for mypy.
* docs: finish the gate doc sweep across README + RAG + generated artifacts
Catch the remaining surfaces beyond the canonical docs.
- README + how-to: agent count 22→25, the 6-agent cells (+ PR Reviewer), the
main reviewer's root→master gate role.
- RAG: permissions + tool-permissions + task-tools list the gate verbs
(claim_gate_review / pr_pass / pr_fail) for pr_reviewer; regenerate the
lifecycle artifacts (intent-verbs, status-transitions, the per-role
lifecycle-*.md prompts, panel lifecycle.json) from the spec via
build_lifecycle_artifacts.py so they carry the new status + verbs.
* fix(migration): shorten the 040 revision id to fit alembic_version VARCHAR(32)
The revision id '040_taskstatus_awaiting_pr_review' is 33 chars; alembic's
alembic_version.version_num column is VARCHAR(32), so recording the migration on
a real 'alembic upgrade head' failed with 'value too long for type character
varying(32)' (surfaced on the NAS deploy). The test suite missed it: the test DB
is built via Base.metadata.create_all and the parity test only renders SQL
offline, so nothing actually applied the migration chain.
- Rename to '040_awaiting_pr_review' (22 chars).
- Add a guard test asserting every revision id fits the VARCHAR(32) column.
- Verified by applying the full chain 001->040 against real Postgres: it now
reaches head and records '040_awaiting_pr_review' without truncation.
* fix(migration): land the actual 040 revision-id shortening + guard test
The prior commit captured only the file rename (git add aborted on the deleted
old path), leaving the long revision id and missing the guard test. This commit
carries the real content: revision id '040_awaiting_pr_review' (22 chars) and the
revision-id length guard. Re-verified against real Postgres — the full chain
reaches head and records the short id without truncation.
* fix(product): flush cell deletes before inserts when re-mapping projects
Editing a product's cell->project map (PATCH /api/products/{id}) 409'd with
'duplicate key value violates unique constraint uq_product_projects_product_team'
whenever a team already had a mapping. _replace_cells clears the old rows and
appends the new ones, but within a single flush SQLAlchemy orders INSERTs before
DELETEs for the same table, so the new (product_id, team) rows collided with the
not-yet-deleted old ones. Flush the deletes first.
Pre-existing bug (unrelated to the PR-review gate); surfaced on the NAS. New
real-Postgres regression test re-maps all three cells to different projects —
it fails with the unique violation without the fix and passes with it. The
existing update test only changed WHICH team was mapped, so it never collided.
* fix(gateway): let main_pm submit_root past the shared submit-up guard
submit_root reused the cell PM's _submit_up_ownership_guard, which
hardcoded agent.role != cell_pm and rejected the Main PM with
"submit_up is reserved for cell_pm". A branch-bearing code root could
then never close: submit_root bounced to complete, while complete
required awaiting_pm_review (reachable only via submit_root) and bounced
back — a circular rejection.
Both callers already run the spec gate (can_invoke_intent), which
enforces submit_up→cell_pm and submit_root→main_pm, so the guard's role
re-check was redundant for submit_up and wrong for submit_root. Broaden
it to accept either PM role as a defense-in-depth non-PM reject.
Adds the first choreographer-level submit_root test (the gap that let
this ship).
* fix(gateway): proactively steer both PMs to their bubble-up verb
The submit_root deadlock had a sibling steering gap: nothing told a PM
which verb opens the gate. The delegate next-hint said only 'i_am_idle
when done', and complete's in_progress rejection named submit_root for
the Main PM but left the Cell PM with a bare 'not ready for completion'
— no submit_up pointer, the same guess-the-verb trap.
- delegate hint now names the role-correct verb (root → submit_root,
cell parent → submit_up) proactively, before any rejection.
- cell_pm_complete's in_progress rejection now steers to submit_up,
mirroring the Main PM's submit_root gate hint.
Tests cover both the cell-PM steer and the role-aware delegate hint.
* docs: correct who-merges-which-PR across the gate docs + complete description
Audit of the gate docs found the merge actors mis-stated in several
places — the exact ambiguity that risks 'the reviewer/PM merges the root
PR' confusion:
- complete IntentSpec description said 'Main PM merges root PR' — false;
main_pm_complete escalates and the CEO merges root→master. Corrected
(propagated to intent-verbs.md, lifecycle.json, generated role prompts
via build_lifecycle_artifacts.py).
- task-tools.md: submit_up target was awaiting_pm_review (should be
awaiting_pr_review); Main PM flow had no submit_root — added it.
- README.md: lifecycle diagram now shows the awaiting_pr_review gate.
- cell-pm.md / main-pm.md: dropped the stale 'submit_up hands work to the
Main PM who merges your cell branch' model — the cell PM merges its own
gated cell→root PR; the Main PM owns the root + submit_root; the CEO
merges master. Added submit_root to the main-pm manifest.
- git-commits.md, pr-creation.md, tool-permissions.md, git-tools.md:
stopped attributing root→master PR opening to complete (it's submit_root).
No behavior change; verb wiring + state machine verified gap-free this
session (the pr_fail→needs_revision→PM respawn loop closes correctly).
* fix(orchestrator): stop closure respawn waiting the reaper window
A PM that finished its subtasks and idled left its parent 'paused' with a
fresh last_heartbeat_at. _is_recently_paused gated closure respawn on
_claim_heartbeat_ttl — the REAPER window (stale_claim_reap_seconds: 600s
default, 1800s on the NAS) — so the parent sat untouched for up to 10-30
minutes before its PM was respawned to close it. The whole chain stalled
behind it.
The race that guard actually protects against (i_am_idle auto-pauses, then
the agent is marked IDLE + its container tears down) is seconds, and the
live-session case is already covered by _is_agent_active. Introduce a
dedicated short debounce (pm_closure_recently_paused_seconds, default 45s)
and gate closure on that instead.
The existing test fixture masked this by setting _claim_heartbeat_ttl to
claim_stale_seconds (180s), not the production reaper value. Fixture now
mirrors production; adds a regression test that a parent paused past the
debounce but within the reaper window respawns immediately.
* feat(gate): post the in-path review verdict on the assembled PR
The in-path gate previously left no trace on the PR it gated — pr_pass /
pr_fail were pure status transitions. Now each verdict is posted as a
GitHub review on the assembled PR itself (server-side, bot account), so
the decision is visible on the very PR the PM merges.
- pr_pass → APPROVE, pr_fail → REQUEST_CHANGES on a cell→root PR.
- The root→master PR ALWAYS gets a plain COMMENT, never APPROVE/REQUEST_
CHANGES: only the CEO acts on master, so the gate must never leave an
approval that could satisfy branch protection (letting someone else
merge) nor a blocking review that could impede the CEO's merge.
- Best-effort and AFTER the DB transition — a GitHub failure is logged,
never rolls back the gate decision. Reuses git.post_pr_review's existing
self-review→COMMENT downgrade for the org's own PRs.
Adds _project_slug_for to the ChoreographerHelpers protocol (mypy) and a
unit suite covering event selection, the master-bound COMMENT rule, the
no-PR skip, and failure-swallowing. Docs updated (pr-reviewer, task-tools).
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
|
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
|
||
|
Renn F
|
f48106cbb6 |
docs: reflow hard-wrapped prose to one line per paragraph
Markdown and editors soft-wrap on their own, so the manual ~75-char line breaks across the docs added nothing but noise. Join wrapped prose, list items, and paragraphs into single lines across 67 docs — README, CLAUDE.md, deployment, usage, the RAG knowledge base, and the agent role prompts. Whitespace-only: code fences, tables, and blockquote alerts are byte-identical and the change is token-verified (no content altered). Applied with a deterministic reflow tool (committed separately). Also lands two doc edits that were awaiting commit: the measured under-load resource numbers in usage.md and the pr_reviewer additions to the org-structure RAG doc. |
||
|
Renn F
|
60de349934 |
fix(prompts): Main PM must honor explicitly-named cells, not silently drop them
A panel-wide UI task that asked for the UX/UI and Frontend cells was
decomposed into Backend + Frontend only — the Main PM judged the UI work
collapsible into Frontend and dropped UX/UI entirely. Its prompt gave full
cell discretion ('you decide which cells', 'most roots only touch one
cell') with no rule to honor cells the brief names.
Add a rule: when the brief, acceptance criteria, or PO/HoM handoff
explicitly call for a cell, the Main PM must create a subtask for each
named cell and never collapse one into a neighbour — its discretion covers
only un-named scope; a genuinely-unnecessary named cell must be confirmed
via escalate_up/dm, not silently dropped. Matching anti-pattern added.
The companion finding (cross-dev sequencing) needed no change: the Cell PM
prompt already keeps dependent units in one dev's lane and splits only
independent units, and cross-cell order is enforced by the dependency_ids
gate, not sequence.
|
||
|
Renn F
|
df5e579916 |
docs: add the full build-session video, count 22 agents, ground resource usage
Add the 2.5-hour 'Working with RoboCo' build session (a conversation to a shipped feature) as a second hero thumbnail beside the 26-min intro. Update the agent count from 20 to 22 across the README, CLAUDE.md, usage, the base agent prompt, the how-to guide, and the org-structure RAG doc: the standing org gains the PR Reviewer (board-level, read-only), and the on-demand Intake and Secretary are now counted. The org-structure doc gains the PR Reviewer in the hierarchy, count table, board team, and communication matrix. The historical 0.1.0 changelog entry is left as-is. Rewrite the resource-usage section: drop the unmeasured per-agent RAM ceiling (RAM is low and agents run few-at-a-time) and lead with storage — the image set's shared base layer — which is what docker prune reclaims. |
||
|
Renn F
|
5bea82dbbc |
fix(gateway): resolve adversarial-review findings on the pr_reviewer flow
An adversarial review of the feature found two blocking defects (both would surface the moment external_pr_enabled is turned on) plus hardening gaps: - HIGH: the enforcement legacy role-gate overlay OVERWROTE spec-derived roles, so pr_reviewer was erased from the (in_progress->completed) edge it shares with the PM self-complete gate — the review task could never complete. Fix: UNION legacy + spec roles instead of overwriting (also preserves the legacy 'add roles' intent on every shared edge). - HIGH: claim_pr_review routed claim+start through the verb runner, which hit start()'s plan gate (planless review task -> None -> crash/respawn loop) and auto-created+pushed a stray branch (violating the read-only/branchless invariant). Fix: mirror QA's claim_review — a verb-body TaskService.pr_review_claim does pending->in_progress with no plan and no branch. - MED: add the pr_reviewer Write(*)/Edit(*) deny at the permission layer (it ingests untrusted PR diffs — make read-only explicit, not implicit). - MED: regenerate the verb-table artifacts (the schemas existed but the generator had not been re-run; the agent prompt showed 'unknown' signatures). ruff + mypy clean (279 files); foundation + gateway suites green (5205 passed). |
||
|
Renn F
|
5902c0fe38 |
feat(roles): add the read-only pr_reviewer role end-to-end
A global, read-only PR reviewer agent (pr-reviewer-1) that reviews inbound external/fork PRs and posts one change-request. Wired end-to-end: - identity: Role.PR_REVIEWER + agent + ROLE_LEVEL (QA-peer) + REVIEWER_ROLES - lifecycle: CLAIM_RULES + ROLE_TEAM_RULES + a dedicated claim_pr_review / post_pr_review verb pair (distinct from QA's) + the pr_review_done action and its in_progress->completed transition; give_me_work / i_am_idle gain the role - role_config: a read-only RoleConfig (allows_write=False) - journaling: ALL_CELLS read tier so it can read internal intent like QA - tracing: post_pr_review requires a learning entry; claim_pr_review is waived - seeds presentation + factory prompt layer + builtin tools + the agentrole enum migration (037) + regenerated verb/lifecycle artifacts Read-only at /app like QA/auditor; default-off — nothing dispatches review work until external_pr_enabled. Foundation + role-config + enum suites green; ruff + mypy clean; orchestrator boots. |
||
|
Renn F
|
92543ad593 |
chore(prompts): stop generating verb tables for driver-based roles
regenerate_verb_tables.py looped every role in ROLE_CONFIGS, emitting a _generated/<role>.md for prompter and secretary too. Both intentionally keep only note+evidence in role_config — their real tools live in their agent_sdk drivers (intake: propose_draft; secretary: read_state/read_task/ submit_directive, the last gated through the backend /directives), and neither uses the _generated/<role>.md prompt-composition path. So the generated tables understated those roles and showed up as perpetually-untracked noise. Skip the driver-based roles (_DRIVER_BASED_ROLES) in both the aggregate verbs.md and the per-role file output, with a comment pointing at the real surfaces. Regenerated verbs.md drops the two misleading sections. |
||
|
Renn F
|
e209e285b8 |
feat(dispatch): per-dev sequenced queues for code subtasks (guardrails spec 3)
True two-dev parallelism: a cell PM delegates the FULL set of code units up front — each dev gets its own queue, both build at the same time, each works its queue one task at a time in order. Replaces the old ceiling (≤2 code subtasks per parent, one per dev) which structurally forced under-decomposition. - Cap: `code` removed from `_SPINE_TYPE_CAPS` — no per-parent code cap (total fan-out still bounded by `_SUBTASK_HARD_CAP=12`); `planning`/`documentation` stay sequential at 1. `_same_assignee_rejection` exempts `code` so a dev may own a queue, but still rejects an exact same-title duplicate (the accidental re-delegation bug). `_spine_type_dup_envelope` simplified to the sequential spine it now only serves. - Dispatch barrier: `_blocked_by_earlier_lane_sibling` holds a dev's higher-sequence pending code leaf while it still has an earlier non-terminal code sibling under the same parent (keyed on assignee, gates only code) — the dev works its queue in order. Wired into `_spawn_pending_dev`. Loop-free (skip the tick, no reject/respawn) and best-effort (lookup failure → dispatch), mirroring the existing merge barrier. The merge barrier is unchanged: leaf PRs still merge serially in sequence order into the shared cell branch, so the independent build lanes never wedge it. - Prompt: cell_pm role guidance rewritten from the two-subtask-cap model to the per-dev-queue model (delegate all units now; dependent units go in one dev's queue, upstream first). Independent per-dev queues (each lane advances at its own pace) rather than strict cross-dev wave-sync, by design — more parallel and leaves the wedge-prone merge barrier untouched. Pairs with the spec-2 idle coverage gate: removing the code cap lets a PM claim every criterion up front, so that gate is always satisfiable. |
||
|
Renn F
|
1fb723174a |
feat(gateway): decomposition coverage gate + AC visibility (guardrails spec 2)
The decomposition floor that pairs with the roll-up gate (spec 4): a PM
cannot finish decomposing a parent while one of its acceptance criteria has
no subtask responsible for it — the "two leaves, half the ACs silently
dropped" pattern. Three parts:
- Gate: i_am_idle is rejected for a cell_pm/main_pm whose owned parent still
has criteria in unclaimed_parent_acceptance_criteria (claimed = referenced
by any live, non-cancelled child). Distinct from the roll-up gate, which
fires at submit_up/complete and demands a *completed* child; this fires
earlier and asks only that every criterion be *claimed*. Safe-by-
construction: inert until a PM declares coverage, so legacy / not-yet-
adopted decompositions are never blocked.
- Visibility: PM-facing briefings (give_me_work, i_will_plan, submit_up) and
every delegate response now carry parent_ac_coverage ({id,text,claimed,
verified} per criterion) + unclaimed_parent_acs, so a PM can map subtasks
to criterion ids via covers_parent_criteria and see what is still
uncovered after each delegate. Off for leaf roles, so a developer's own
criteria never surface as bogus "unclaimed" noise.
- Prompts: cell_pm / main_pm role prompts document covers_parent_criteria and
the new idle enforcement in the existing Coverage discipline.
TaskService.{parent_ac_coverage,unclaimed_parent_acceptance_criteria} added
beside uncovered_parent_acceptance_criteria; all three refactored onto a
shared _parent_ac_ref_sets helper (keeps each under the xenon B ceiling,
preserves the committed roll-up behavior). Verb tables regenerated for the
new delegate param — the regen also syncs pre-existing table drift that was
never regenerated after earlier merges (read_messages, pass_review
ac_verdicts, board pitch). Two brand-new generated tables (prompter,
secretary) are left untracked pending a separate decision.
|
||
|
|
46d89b58fe |
feat: company-in-a-box — goal-aware company layer (0.4.0) (#171)
* feat(goals): company charter singleton — data layer (Business Goals slice 1)
First slice of the company-in-a-box "Business Goals" phase: a single CEO-owned
charter row (north star + objectives + constraints + operating policy) that
will be injected into every agent's context_briefing so all work is goal-aware.
- CompanyGoalsTable: singleton table (all-zeros id), JSON objectives /
constraints / operating_policy, updated_at / updated_by.
- migration 032: create + seed the singleton row (offline-renderable; column
server-defaults fill an INSERT of just the id).
- CompanyGoalsService: get() (empty defaults when unset) + upsert() (singleton,
partial update, caller commits).
- tests: empty defaults, roundtrip, singleton + partial-update preservation.
Next slices (mapped, not yet built): briefing injection (BriefingInputs +
build_context_briefing + EvidenceRepo), API route (GET any / PUT CEO-only),
panel /goals page, and base/Board/PM prompt mentions.
* feat(goals): inject the company charter into every agent briefing (slice 2)
The charter is now goal-aware context for every agent:
- BriefingInputs gains company_goals; build_context_briefing surfaces it.
- EvidenceRepo.company_goals(): single-row lookup returning a COMPACT charter
(north star + objectives + constraints + operating policy; audit columns
dropped, lists capped) or None when unset, so an empty charter never bloats
the per-verb briefing.
- _briefing_for wires it into every context_briefing.
Tests: briefing surfaces company_goals (defaults None); repo returns None for an
absent/empty charter and the compact dict when set.
* feat(goals): company charter API — GET any agent, PUT CEO-only (slice 3)
- routes/company_goals.py: GET returns the charter (any authenticated agent —
it drives every briefing); PUT is CEO-only (403 otherwise), partial update via
model_dump(exclude_unset=True), explicit commit.
- schemas/company_goals.py: response + partial-update models.
- registered at /api/company-goals.
- tests: GET open to any role, CEO update persists + is readable, non-CEO 403.
* feat(goals): make the company charter actionable in agent prompts (slice 5)
Agents already receive company_goals in the briefing (slice 2); now tell them to
act on it:
- base.md: universal "Align with the company charter" section — favour work and
trade-offs that advance the objectives, honour the constraints, flag conflicts;
never a license to leave your role.
- board / main_pm / cell_pm: role-specific lines tying triage / cell-routing /
subtask decomposition to the charter.
Prompts are composed at spawn from base.md + roles/*.md directly (compose_prompt),
so no _generated regeneration is needed.
* feat(goals): company charter panel page (slice 4)
CEO-facing editor for the charter at /company-goals:
- lib/api/company-goals.ts: get / update (PUT) client.
- company-goals-card.tsx: edit north star + constraints (one per line) +
objectives / operating_policy (JSON, parsed + validated with toast errors);
display derives from server state (no set-state-in-effect).
- (dashboard)/company-goals/page.tsx + a "Company Goals" sidebar nav link.
tsc --noEmit + eslint clean. Completes Phase 1 (Business Goals): data, briefing
injection, API, prompts, panel.
* fix(test): make test_app route assertions robust to FastAPI 0.137 _IncludedRouter
FastAPI 0.137 stopped flattening include_router into app.routes — each include is
now an _IncludedRouter (a BaseRoute with no .path), so `{r.path for r in
app.routes}` raised AttributeError and the two router-registration tests failed
(the bump arrived via the claude-agent-sdk update in uv.lock). Add
_registered_paths(): OpenAPI schema paths (the stable public contract) plus each
included router's prefix, which also covers the websocket /ws mount (never in the
schema). Drops the now-incorrect type: ignore[attr-defined].
* feat(research): pluggable web search/fetch for Board + PM agents
Add a provider-agnostic web-research capability so the Board and PMs can
ground decisions in current external evidence the knowledge base can't
answer.
- ResearchService selects a provider adapter from config: Tavily, Brave,
and Exa adapters plus a NullProvider that degrades gracefully when no
key is set. Result count and fetched-content size are clamped to caps.
- /api/research/search and /api/research/fetch: role-gated to Board + PMs
(and the CEO), with a per-agent/day Redis quota that fails open.
- roboco-search MCP server (web_search / web_fetch) calls those routes;
the provider key stays server-side and agent containers never egress.
Mounted per role by the orchestrator, behind a master switch.
- Charter-aware prompt guidance for Board, Main PM, and Cell PM.
Additive: with no key configured it is a no-op and the existing delivery
lifecycle is unchanged.
* feat(pitch): Board pitch -> CEO approve -> auto-provision repos
Add an additive origination path so a product can be proposed, approved,
and stood up without manual repo/Project setup.
- Pitch entity + migration (pitches table); PitchService create/list/
reject/approve.
- GitHubProvisioningService: the one place that creates repos (POST
/orgs/{org}/repos). Server-side token/org; when unconfigured the whole
approve path is inert and nothing is created.
- On approval: provision one repo per target cell, register a Project per
repo, create a Product when multi-cell, and seed one Main-PM delivery
task — all reusing the existing Product / coordination-task machinery.
- /api/pitches: Board authors (PO/HoM), CEO approves/rejects, Board+PM+CEO
view. Errors mapped via a single translator.
Additive: the delivery lifecycle is untouched; with no provisioning token
the capability is a no-op. Agent-facing pitch tool + panel are follow-ups.
* feat(strategy): dormant autonomous strategy engine (engine 2)
Add a second, optional engine that watches the company against its
standing goals and surfaces what needs the CEO — without touching the
delivery lifecycle (engine 1).
- StrategyEngine.assess() reports observations: the company is idle while
goals stand, and tasks stranded in 'blocked' past a threshold.
- run_cycle() notifies the CEO (notify-only; it never spends, builds, or
auto-approves — originating work stays a CEO decision).
- Orchestrator runs it on its own interval, started/stopped with the other
background loops; the loop returns immediately unless enabled.
DORMANT by default (strategy_engine_enabled=False): the loop never runs and
a standard deployment is unchanged. Auto-origination is a further opt-in.
* docs(changelog): record Business Goals, Web Research, Pitch->Provision, and the dormant strategy engine under Unreleased
* feat(secretary): wire the Secretary role end-to-end (foundation)
Add SECRETARY as a distinct role — the CEO's conversational chief-of-staff,
governed separately from the Prompter (which stays read-only/human-only).
This is the role foundation only; authority, the live agent, and the panel
land in following commits.
- foundation/identity: Role.SECRETARY (board level), seeded secretary-1 agent,
role-level mapping.
- journaling read tier (ALL — it advises the CEO), role_config entry,
per-role model (opus), prompt-layer mapping + roles/secretary.md.
- i_am_idle gains SECRETARY so the role has a verb surface.
- migration 034: add 'secretary' to the agentrole enum (mirrors 025).
- Role-registry tests updated for the new role.
Inert by itself (nothing spawns it yet); additive — existing roles unchanged.
* feat(secretary): directives + gate-list authority (backend)
The Secretary acts only under CEO command. Low-risk directives (relay a
dictated message) execute immediately; high-impact ones — charter edits,
task start/cancel/override, pitch approval, announcements — are recorded
pending and run only after the CEO confirms (the gate list).
- secretary_directives table (migration 035) as the command audit + queue.
- SecretaryService: read company state; submit (direct->run, gated->queue +
notify CEO); confirm/reject; execution runs with the CEO as actor through
the existing services (the Secretary never holds CEO authority itself).
- /api/secretary: submit + state/task reads (Secretary or CEO); list/confirm/
reject (CEO only). Writes commit explicitly.
* feat(secretary): live conversational agent (container + bridge)
Stand up the Secretary as a persistent Claude-SDK container the CEO chats
with, mirroring the Intake agent and reusing its driver/session machinery.
- secretary_driver: build_secretary_options exposes read_company_state /
read_task / submit_directive as SDK tools that call /api/secretary/* with
the agent's HMAC token; backend-call logic is module-level + tested.
- secretary_main: container entrypoint (receiver + relay) reusing IntakeDriver.
- orchestrator: start/spawn/reap secretary session + run-cmd builder; no
workspace clone (reads state via API), mints a role=secretary token.
- secretary_live routes: panel <-> container bridge over the live registry.
- agent-secretary image (Dockerfile + compose build service).
Inert until a session is started; additive — intake and all agents unchanged.
* feat(secretary): panel chat + directive confirmation queue
The CEO's Secretary surface: a live chat (SSE) to talk to the Secretary, and
a 'Needs your confirmation' queue listing gated directives the Secretary
proposed — each with Confirm / Reject. Adds the sidebar nav entry.
- lib/api/secretary.ts: live (start/stream/status/send/stop) + directive
(list/confirm/reject) + state clients (all as the CEO).
- hooks/use-secretary.ts: drives one chat, accumulating SSE token deltas.
- secretary page: chat pane + pending-directive cards.
Completes the Secretary end-to-end (role + authority + live agent + panel).
* feat(pitch): agent-facing pitch tool + pitches panel
Complete the pitch path: the Board can now author pitches through the gateway,
and the CEO reviews/approves them in the panel.
- content_actions.pitch (Board-only) -> PitchService.create, returning an
Envelope; wired as a do-tool (do_server + /api/v1/do/pitch + schema) and
added to the Board's do-tools.
- Panel /pitches page: lists pitches with CEO Approve & provision / Reject;
sidebar nav entry.
Pitch (Phase 4) is now end-to-end: author -> CEO approve -> auto-provision.
* feat(cockpit): read-only 'is the business winning?' summary
A pure aggregation for the CEO over existing data — no new state, no writes.
- CockpitService.summary(): charter north-star/objectives, delivery counts
(in-flight/blocked/awaiting-CEO), 30-day spend vs the charter's budget cap,
pending pitches, and the strategy engine's signals (what needs you). Stamped
basis='proxy' — performance is a proxy until real launches.
- GET /api/cockpit/summary (CEO / Board / Main PM / Secretary).
- Panel /cockpit page + sidebar nav.
Reuses goals + usage + StrategyEngine.assess(); reads only.
* docs(changelog): add the Secretary and Cockpit to Unreleased
* fix(test): isolate the company-goals empty-defaults test from committed state
The shared test DB persists committed writes across tests; a route test
commits a charter, so the unit test's 'unset' assertion must establish its
own clean precondition rather than assume global emptiness.
* fix(gateway): lower evidence_repo complexity to rank A (xenon gate)
company_goals()'s 4-way `or` emptiness check tipped the module average to
rank B; `any(...)` is equivalent and keeps the module under the gate's A bar.
* chore(compose): mirror agent-secretary-image build into docker-compose.yaml
Both compose files are byte-identical and tracked; .yaml carries the same
agent-secretary-image build service already present in docker-compose.yml.
* chore(lifecycle): regenerate artifacts for secretary i_am_idle
The secretary role gained i_am_idle in the lifecycle spec; regenerate the
generated prompt/doc/json artifacts so foundation-check stays green.
* docs(changelog): cut the company-in-a-box phases to 0.4.0
Label the six additive phases (business goals, web research, pitch-provision,
strategy engine, secretary, cockpit) as 0.4.0; tag v0.4.0 is held until the
branch merges to master so it points at the release commit.
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
Renn F
|
481c5fe17b |
feat(gateway): run a fast quality gate at i_am_done, before QA
The developer's i_am_done submit now runs the project's fast quality gate (lint + typecheck) in the developer's workspace and blocks the transition to awaiting_qa if it's red, returning the failing output as the remediate hint — so a red gate is caught at the dev's desk instead of in QA review or CI. The slow test suite intentionally stays on CI. The gate is fail-open on infrastructure errors (missing workspace/toolchain never blocks a submit) and a no-op for projects that configure no lint/typecheck commands. Developer prompt updated. |
||
|
Renn F
|
2db11c7833 |
feat(qa): require a per-acceptance-criterion verdict before pass_review
QA may no longer pass a task with a single gestalt approval — pass_review now takes ac_verdicts (one verification entry per acceptance criterion) and the gateway rejects a pass that does not cover every criterion. If a criterion does not hold, QA fails the review instead. The verdicts are folded into the persisted qa_notes for the audit trail. Threaded through the flow MCP tool, the HTTP request schema, and the route; QA prompt updated. |
||
|
Renn F
|
8affb283f5 |
feat(gateway): enable 2-devs-per-cell parallelism + split-before-claim sizing
- Intake / Main-PM / Cell-PM prompts: enumerate independently-shippable work units, inherit the breakdown down the chain, and dispatch independents in parallel (dependency order, never one-at-a-time). - Raise the code-spine concurrency cap from 1 to 2 per parent (one per cell developer) so both devs build in parallel; keep the same-assignee guard and the planning/documentation cap at 1, plus the cross-team planning exemption. - Split-before-claim: hard-block an egregiously-bundled code leaf at delegate time so the PM splits it before any dev claims it; nudge the moderate band in the delegate success envelope. |
||
|
|
a6b67a6a58 |
Feat: board redraft loop (#139)
* feat(board): expose board review brief + guard approve-and-start
Slice 1 of the board-informed intake re-draft loop (backend foundation):
- JournalService.board_review_brief(task_id): the PO + Head of Marketing
DECISION_LOG entries for a task, oldest-first, each tagged with author —
the board's review as structured data.
- GET /api/tasks/{task_id}/board-review (PM-or-above) backing the CEO's
approval/redraft surface, so the real board analysis is readable instead
of a placeholder; BoardReviewEntry response schema.
- Guard: approve_and_start now refuses a board task whose review is not
complete (service invariant + precise BOARD_REVIEW_INCOMPLETE at the route).
Previously only the UI hid the button; the backend let an early/rogue call
hand the task to Main PM mid-review.
Tests: brief filtering/ordering + endpoint (200/404) + the two guard paths.
* feat(panel): show real board review at the approve gate + live refresh
Slice 1 frontend of the board-informed intake re-draft loop:
- tasksApi.getBoardReview + useBoardReview hook consume GET
/tasks/{id}/board-review.
- The Approve & Start dialog now renders the actual Product Owner + Head of
Marketing notes (markdown) instead of a static placeholder, so the CEO reads
the board's analysis before approving.
- C2: useTask polls (4s) while a task is still on the board with an
incomplete review, so the Approve & Start button appears as soon as the
board finishes — there is no per-task websocket. Polling stops once
board_review_complete flips.
* feat(intake): board-informed re-draft loop (backend, cold path)
Slice 2 of the re-draft loop:
- update_live_draft: apply a board-informed re-draft to the EXISTING task in
place (title/description/acceptance_criteria) — never a duplicate — then route
it: 'main_pm' hands it to the Main PM via approve_and_start; 'board' clears
board_review_complete for another review round.
- confirm route branches on task_id → update_live_draft vs confirm_live_draft;
LiveConfirmRequest.task_id added (scope taken from the task, not required).
- POST /live/re-interview/{task_id} (PM-or-above): spawns a fresh intake session
seeded with the current draft + the board brief (compose_redraft_message),
scoped to the task's product/project. The cold path + Slice-3 fallback.
- format_board_briefing / compose_redraft_message helpers.
Tests: pure helpers + update_live_draft (main_pm hand-off, re-board reset,
missing-task).
* feat(panel): board-informed re-draft entry + prompter re-draft guidance
Slice 2 panel of the re-draft loop:
- 'Re-draft with board feedback' button on a board-reviewed task detail →
/prompter?redraft=<taskId>.
- usePrompter.startRedraft(taskId): calls POST /prompter/live/re-interview/{id},
scopes the chat to the task, and streams the re-draft; redraftTaskId is
threaded (persisted across reload) so confirm carries task_id and updates the
existing task in place rather than creating a duplicate.
- prompterLiveApi.reInterview; ConfirmPayload.task_id.
- Prompter role prompt: a 'Re-drafting after board review' section so the agent
revises the included draft from the board brief instead of starting over.
Panel verified by CI (no local node_modules).
* feat(intake): keep-alive re-draft — park the intake agent during board review
Slice 3 of the re-draft loop (in-context fidelity; cold path is the fallback):
- Registry: LiveIntakeSession.task_id + park(session_id, task_id) (keep alive
instead of reaping) + find_by_task() for board-completion injection.
- Confirm: the board route (first pass) PARKS the intake agent instead of
reaping, so it keeps the whole interview in context.
- Orchestrator: on board-review completion, inject the synthesized board brief
into the parked session (_inject_board_brief_into_parked_intake) so the
resident prompter re-drafts in-context. No-op when nothing is parked (the
container died / a new intake replaced it) — the cold /re-interview path
covers that. No reaper change needed (an idle parked session spends no tokens
and the budget sweep is the only agent-stopping sweep).
- Panel: confirm(board) keeps the chat alive (parked, redraftTaskId set) with a
notice; the injected revised draft arrives over the existing stream to approve.
Tests: registry park/find_by_task/closed-ignored. Container delivery + the full
panel parked flow need live (container-runtime) verification.
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
Renn F
|
e0cd305844 |
docs(cell-pm): require mapping every cell criterion to a subtask before idling
Decomposition is where scope silently disappears: a cell PM delegates a subtask covering most of its acceptance criteria, idles, and the uncovered criteria have no subtask, no dev, and no branch — the gap surfaces only at submit_up or, worse, at QA/CEO review, forcing a full cell revision loop. Add a Coverage section to the cell-PM prompt that pulls the every-criterion- has-a-home discipline forward from the submit_up checklist to decomposition time. Before idling after a delegate, the PM must account for every cell criterion as one of exactly three outcomes — covered now, covered later in a sequenced follow-on (recorded in the decision note), or out of cell scope (also recorded) — so a dropped criterion costs one extra delegate instead of a whole revision loop. Reinforces that on respawn the anti-re-decompose rules will block recovering dropped scope, so coverage must be mapped up front. |
||
|
|
b9082a5c70 |
Fix: agent idle deadlock and lifecycle hardening (#96)
* fix(panel): cap dialog height and pin footer so actions stay reachable
Shared DialogContent now caps at max-h-[85vh] with overflow-y-auto, and the
footer is sticky to the bottom. Long content (e.g. a pasted change-request
note) no longer pushes the submit/cancel buttons past the viewport — the body
scrolls while the actions stay visible. No-op on dialogs that already fit.
* feat(notifications): suppress duplicate same-purpose notifications at send
A notification is not created when an unacknowledged one with the same purpose
— same sender, same type, same task, overlapping recipients — already exists.
Body text is not compared, so rewording cannot defeat it; a different type,
task, sender, or an already-acked recipient all still send through. Stops
agents that loop re-issuing the same signal from piling up unread that
soft-blocks the recipient's idle path.
* fix(gateway): stop board/PM lifecycle verbs from 500-crashing
Two unguarded crashes that wedged the org in respawn/escalate loops:
- escalate_to_ceo dereferenced None.status when the verb runner declined the
escalation (task not in awaiting_pm_review — e.g. a board agent escalating a
blocked task). It now returns a clean invalid_state. The message/remediate
build moved to a helper so the function stays within the complexity gate.
- The coordination-root git ops (pr_target, pr_merge, PR update, branch-token
resolve) called UUID(str(task.project_id)) directly, which raised on a
coordination/integration task (project_id is None — 'badly formed hexadecimal
UUID string'). They now resolve through _project_for_task, which falls back to
the product's repo for project-less roots.
* refactor(intake): split out _block_to_chunk per-block classifier
Extract the per-block classification from _blocks_to_chunks so each function
stays within the xenon cyclomatic-complexity gate (was rank C). Behaviour is
unchanged — verified by the existing intake_driver tests.
* feat(gateway): make the i_am_idle unread soft-block satisfiable
The soft-block on unread A2A / @mentions had no clearing path, so once those
briefing fields populated an agent could never idle — a whole-org deadlock.
Keep the guard (it is correct) and add the missing clear paths:
- New read_messages content verb (schema -> route -> handler ->
a2a.mark_all_read -> MCP tool -> role do_tools): bulk-zeroes the caller's
unread A2A and stamps read_at. The idle hint now points to it.
- list_unread_mentions returns UNACKED MENTION-type notifications (each @mention
already raises one via messaging._notify_mentions) instead of raw,
unconditional mentions, so they clear via the existing notify_ack. No schema
migration needed.
The soft-block is now satisfiable: A2A via read_messages, mentions and
notifications via notify_ack.
* fix(tests): repair notification-dedup db.scalar mocks + prompter agent seeding
The notification send-dedup added a db.scalar() purpose-lookup to
_create_notification; the two hand-rolled _FakeDb test stubs (test_notification,
test_a2a_priority_tristate) had no scalar() method → AttributeError. Add
scalar() returning None (no duplicate) so creation proceeds.
Separately, the prompter '& Start' route tests assign the draft to a fixed
product-owner / main-pm AGENT_UUID but only seeded system + CEO, so the
assigned_to FK failed in isolation (and main-pm flaked in the full suite). Seed
both via idempotent merge() in _seed_project_and_ceo.
* fix(git): gitignore .pnpm-store + flag GH001 push rejection as permanent
A dev once committed the ~115 MB pnpm store → GitHub GH001 (>100 MB) pre-receive
reject → open_pr retry-loop. Two root fixes:
- Add .pnpm-store/ to .gitignore — an ignored dir can't be staged by any git add.
- push() restates a GH001 / file-size rejection as an unmistakable PERMANENT
error pointing at i_am_blocked, so the agent stops blind-retrying a push that
can never succeed (it otherwise mis-reads the raw output as a transient timeout).
The per-verb retry cap (open_pr: 5) already bounded the burn; this ends it.
* fix(gateway): accept a PM decision note as satisfying the complete/submit_up reflect gate
A cell/main PM that wrote a fresh decision but no separate reflect note bounced
on the reflect tracing-gate indefinitely (re-confirmed live: cell PMs looped on
cell_pm_complete -> journal:reflect until reaped, burning tokens — worse because
each respawn resets the per-verb retry cap). For a PM closing/submitting a task
the decision note already documents the close; the separate reflect is the
redundant artifact weak-model PMs forget. Accept a fresh decision as satisfying
reflect for complete + submit_up — the gate still requires a decision +
substantive notes, so the close stays documented.
NOTE (enforcement tradeoff, flagged for CEO review): this intentionally relaxes
the PM complete/submit_up gate. It does NOT touch the developer i_am_done gate.
* feat(gateway): refuse i_am_idle when a PM still owns a task awaiting its review
A cell/main PM once tried to 'send work back' by DMing the developer and going
idle — but a DM changes no task state, so the task stayed awaiting_pm_review and
the orchestrator just re-dispatched the PM in a loop. i_am_idle now refuses (like
the pending-assignment guard) when a PM owns an awaiting_pm_review task, with a
clear remediation: complete() to finish, or reassign()/delegate() to route it
back. PM-only; devs/QA/doc unaffected. Pairs with the reflect-gate relaxation so
the PM can actually complete instead of looping.
* feat(gateway): push a prior-work handoff digest into task-scoped briefings
A freshly spawned or respawned agent previously started cold on every
lifecycle hand-off: the prior worker's PR, commits, acceptance status and
journal highlights lived in task evidence but were pull-on-demand, so each
new role agent re-explored the codebase from scratch — wasted tokens and
fragile context loss across respawns.
build_task_handoff() composes a compact, DB-only digest (no git diff) and
_briefing_for() now attaches it to context_briefing whenever the caller
already holds the task row. The digest is built only from a passed-in task,
so there are zero extra fetches: every resumption entry point (give_me_work
and pm_give_me_work, i_will_work_on, i_will_plan, triage/triage_all,
i_am_done, submit_up, escalate_up, complete) threads the loaded task, while
id-only correction/rejection paths cleanly omit it.
Every field is type-guarded so a partial row never leaks a non-serialisable
value into the envelope.
* docs(prompts): tell agents to resume from the briefing handoff before re-exploring
The base prompt described the success envelope but never told agents to act
on context_briefing, so a respawned or hand-off agent would re-scan the
whole repo and re-derive the plan even when the briefing already carried the
prior worker's PR, commits, acceptance status and journal highlights.
Adds a 'Resume from your briefing' section that walks each task_handoff
field and instructs the agent to continue from it — and to read the unread
A2A / mention / notification lists, which are messages addressed to them.
Pairs with the gateway change that now pushes task_handoff into every
task-scoped briefing.
* feat(tasks): remember cleared dependencies so the unblock briefing can surface them
When an upstream dependency completed, _unblock_dependents removed its id from
the dependent's dependency_ids to let it be claimed — destroying the only
record of which upstream task had just landed. The revived dependent then
re-discovered that work from cold.
Adds tasks.completed_dependency_ids (Alembic 026, uuid[] default '{}'):
_unblock_dependents now appends the cleared id there instead of only dropping
it, and the briefing handoff digest surfaces it so the agent picking the task
back up knows its blocker cleared because that upstream work shipped. The base
prompt documents the field.
Migration round-trip verified against postgres (upgrade adds the column,
downgrade drops it).
* docs(prompts): instruct PMs to split oversized tasks into per-concern subtasks
A subtask carrying a long acceptance list or spanning multiple layers/files
drove repeated QA failures and a PM revision loop — QA can't pass a partial,
and the dev keeps re-touching unrelated parts. Nothing in the PM prompts told
them to decompose by size/concern.
cell_pm gets a 'Sizing' rule: one subtask = one focused concern with ~2-4
criteria and its own dev->QA pass; decompose anything larger before
delegating, sequencing with dependencies. main_pm gets a matching reminder to
scope each cell's slice to that cell's layer rather than handing a cell a
cross-layer monolith that just pushes the problem down a level.
* fix(gateway): mirror the task= kwarg on ChoreographerHelpers helper signatures
The handoff-digest change added a keyword-only task= parameter to
_briefing_for and _build_tracing_gap in _impl, but the ChoreographerHelpers
base that the role mixins inherit still declared the old signatures, so the
composed Choreographer had two incompatible base definitions (mypy [misc]).
Sync the base declarations to match.
* fix(tasks): keep the owner on a substitute-out so the task isn't orphaned
build_substitute_update unconditionally nulled assigned_to, so any
substitute that routes to PENDING (max_retries, low_context, out_of_scope_*)
— the path a verb hitting repeated 500s or its retry limit takes — left the
task pending AND unassigned. The dispatcher only respawns a pending task when
it has an owner, so the task went dormant: no agent ever picked it back up.
Keep the task with its current owner instead. A substitute-out is almost
always a transient stall, so the task re-dispatches to the SAME agent, which
resumes from the briefing handoff. Only the task_complete -> PM-review handoff
changes owner (unchanged).
* feat(a2a): suppress duplicate unread A2A messages at send
A respawned or retrying agent could re-emit the same DM, stacking identical
copies on the recipient's inbox and re-bumping the unread count — noise that
the recipient then has to clear. The notification path already dedups; A2A did
not.
send_chat_message now suppresses a send when an identical message from the
same sender is still unread in the conversation, keyed on (conversation,
sender, message_kind, content). Genuinely different messages are never
collapsed (verified: distinct content still produces distinct rows), so this
avoids the earlier per-pair over-suppression. No migration.
* fix(panel): default the notifications view to Unread, not All
Landing on the All tab buried new notifications under everything already
seen — the most-reported annoyance. The Unread tab is the actionable view, so
make it the default; the All/Pending tabs are one click away.
* fix(panel): show clone progress during intake prep instead of a frozen pill
The first clone of a repo can take a few minutes, during which the intake
form showed only a static 'Preparing the agent…' button — indistinguishable
from a hang. Add a progress region while preparing: an elapsed timer, a
saturating progress bar (approaches but never reaches 100% until the agent
actually answers), and staged copy (spinning up → cloning → first-clone-takes-
a-while → reading the codebase) so the wait reads as work, not a freeze.
* feat(docs): index workspace-authored docs that never reached the RAG store
Docs written through roboco_docs_write land at /app/docs on the orchestrator
and index fine. But a documenter can also write docs with Edit/Write directly
in its own clone (README, CHANGELOG, workspace markdown); those resolve to a
/app/docs path that doesn't exist on the orchestrator, so the indexer reads
nothing and the docs never become searchable — a cross-container miss with no
shared mount to bridge it.
On docs completion, capture each listed doc's committed content out of the
branch (new GitService.read_file_at_branch, via git show) and write it
server-side under /app/docs before indexing, so workspace-authored docs reach
RAG too. Docs already present server-side are skipped; absolute paths and
unreadable/uncommitted files are passed over best-effort.
* feat(prompter): survive a browser reload by reconnecting to the live intake chat
The intake chat lived entirely in React state, so a page reload wiped it and
dropped the human back to the scope form — even though the agent container
outlives the page. Now the chat persists a small TTL'd slice (session id,
messages, scope, draft) to localStorage and, on mount, reconnects: it asks the
new GET /live/{id}/status whether the session is still running and, if so,
restores the history and reopens the SSE stream; if dead or expired it clears
and shows the form. A full reload doesn't run React effect cleanup, so the
navigate-away reap never fires on refresh and the session stays up.
Backend adds the status endpoint + PrompterLiveRegistry.is_alive; localStorage
is cleared on confirm, start-another, and SPA navigate-away.
* chore: remove internal session-bookkeeping refs from code comments (part 1)
Strip leaked task/finding numbers, Wave/Phase/cluster/audit labels from
docstrings and comments across services, foundation policy, runtime, mcp,
api schemas, and agent_sdk — they mean nothing to a repo reader and expose
process internals. Wording preserved; only the labels dropped. Done by hand,
one comment at a time (no scripted rewrite). _impl.py follows separately.
* chore: remove internal session-bookkeeping refs from code comments (part 2)
Finishes the manual scrub: the choreographer _impl.py docstrings/comments plus
the remaining dogfood-run ('smoke-N') labels across runtime, mcp, foundation,
api schemas, services, and agent factories. Reworded to describe the bug or
behaviour in plain words; every label dropped. The repo source is now free of
task/finding numbers, Wave/Phase/cluster/audit/smoke labels. By hand, one
comment at a time.
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
|
9f8834155a |
Feature: prompter gold upgrade (#84)
* feat(prompter): make the assistant a RoboCo insider and fully wire launch
The Prompter's intelligence lived in two thin static prompts, so it asked
generic checklist questions and produced a flat task. The launch path was
also only half-wired: the panel called the generic task-create endpoint with
no project, bypassing the Prompter's own confirm flow.
Interview brain
- Rewrite the chat system prompt with RoboCo's org model, the task-spec
standard, a dimensions playbook, and a reflect-back, 1-2-questions-per-turn,
auto-stop discipline.
- Inject the live projects/products list each turn so the assistant grounds
questions in real surfaces and resolves the target itself.
- Replace the brittle phrase-match readiness with a parsed roboco-meta control
block (parse_readiness); the block is stripped from the visible reply and the
turn now returns draft_ready + scale.
Structured GOLD draft
- Add first-class draft fields (objective, what_this_builds, the_work, notes)
carried in the existing draft_data JSONB — no migration.
- Compose the GOLD markdown description deterministically from those fields
(compose_description); the model never hand-formats the body.
Adaptive routing + wired launch
- Confirm now runs through the Prompter confirm endpoint with the human's
project/product choice and edited structured draft.
- Single-cell targets a project and the cell team; a multi-cell feature targets
a product and becomes a Main-PM coordination root that fans out.
Frontend
- Turn-envelope draft_ready (drop the duplicated phrase-match), structured
draft card, confirm dialog with a project/product picker and a per-cell
The Work editor, and the corrected priority labels (0 highest .. 3 lowest).
* fix(prompter): commit session writes so they survive across requests
Session create returned 201 but the row was never durably committed, so the
immediately-following /messages call could not find it and 404'd. The prompter
routes were the only write surface that never called db.commit() — every other
write route (tasks, a2a, groups, docs, product) commits explicitly rather than
rely on the request-teardown auto-commit, which is sensitive to middleware and
teardown ordering under the production server.
- Commit explicitly in all four prompter write routes (create session, send
message, get/generate draft, confirm).
- Fix _get_session's NotFoundError: it passed a full sentence as resource_type,
producing the doubled "... not found not found" message; now uses the
(resource_type, resource_id) signature.
- Panel: when a message hits a session the server no longer has, start a fresh
session and retry once instead of dead-ending on a stale id.
Add a regression test that gives each request its own non-committing session —
the real cross-request boundary the shared-session integration tests never
crossed. It reproduces the production 404 without the route commit and passes
with it.
* refactor(prompter): drop the "GOLD" jargon for plain wording
"GOLD" was informal shorthand for "a good/well-formed spec" that should never
have been baked into the LLM prompts, comments, and docstrings as if it were a
defined term. Replace it everywhere with plain language ("a well-formed task",
"a complete task spec", "the markdown description", "structured spec fields").
No behaviour change.
* feat(intake): add the intake interviewer agent role (static definition)
Phase 1 of the intake-agent feature: a new first-class `prompter` role — the
intake interviewer the CEO chats with to draft a task. This commit defines the
role across every foundation layer (no runtime yet); spawning + the live
session come next.
- identity: Role.PROMPTER, RoleLevel.INTAKE (lowest authority), an AGENTS row
(intake-1) on the board team, ROLE_LEVEL entry. Deliberately NOT in
BOARD_ROLES — it interviews, it does not review.
- lifecycle: gets i_am_idle like every agent (its only verb); no
delivery-lifecycle intents.
- journaling: ReadTier.OWN — isolated, reads only its own journal.
- role_config: human-only manifest — note + evidence only, no say/dm/notify/
channels; allows_subagent=True (research), allows_write=False.
- agents_config derives it automatically and correctly excludes it from
TASK_CREATOR_ROLES (it drafts, it never creates tasks).
- seed presentation ("Intake"); regenerated lifecycle artifacts.
- role system prompt: read the code first, single-CEO awareness, propose
rather than interrogate — written against the failures we saw.
- docs: roster count 19 -> 20, org charts, verb-surface table, usage roster.
All foundation drift checks pass; role/manifest/permission tests green.
* feat(intake): migrate agentrole enum to add 'prompter'
ALTER TYPE agentrole ADD VALUE IF NOT EXISTS 'prompter' so the intake agent
row seeds/spawns against a migrated production DB. Forward-only (postgres
can't drop enum values), guarded for offline mode — matches migration 012.
* style(intake): ruff format the role additions
* fix(intake): unguard the agentrole migration so it renders offline
The enum-migration-parity test renders 'alembic upgrade head --sql' (offline)
and greps for ALTER TYPE ... ADD VALUE. The is_offline_mode() guard skipped
emitting it, so the parity check couldn't see 'prompter'. Drop the guard —
PG16 permits ADD VALUE in a transaction, same as migration 020's backfill.
* feat(intake): the live-session driver (Claude Agent SDK loop)
Phase 2 begins. The intake agent isn't a one-shot `claude -p`; it's a live
Claude Code session the human chats with. This driver is the container's loop:
open one long-lived claude-agent-sdk ClaudeSDKClient, then per human message
run a turn (query + receive_response) and stream its events out, keeping
conversation context in-process — verified against the real SDK (v0.2.94).
- StreamChunk + normalize(): map SDK messages (StreamEvent text deltas,
AssistantMessage text/thinking/tool_use blocks, ResultMessage→session_id) to
panel-facing chunks. Duck-typed, so it works on real SDK objects and on test
fakes alike — SDK-free, fully unit-tested.
- IntakeDriver.run(): the loop, with injected session/source/sink seams; a turn
failure surfaces as an error chunk without killing the session.
- SdkIntakeSession + build_intake_options: the only SDK-coupled code (lazy
import; needs the live claude binary, so excluded from coverage).
- Add claude-agent-sdk dependency + mypy ignore-missing-stubs.
Relay, panel SSE, and the persistent on-demand spawn are the next steps.
* Updated uv.lock
* feat(intake): the panel<->agent live bridge (registry, routes, entrypoint, image)
Wires the live intake chat end to end (Phase 2 integration layer):
- prompter_live.py: the orchestrator-side per-session registry — open/close,
push (agent->panel), stream (SSE drain), deliver (panel->container). In-process
(the orchestrator is single-process). 7 unit tests.
- routes/prompter_live.py: GET /live/{id}/stream (SSE), POST /live/{id}/messages
(deliver), POST /live/{id}/events (relay in); registered under /api/prompter.
5 integration tests.
- agent_sdk/intake_main.py: the container entrypoint — a POST /turn receiver
(the driver's MessageSource) + a relay-poster EventSink + the ClaudeSDKClient
session, run concurrently. 4 unit tests on the wiring helpers.
- docker/agent-prompter.Dockerfile: FROM base, ENTRYPOINT = the driver (not the
one-shot `claude` the other agents use).
Remaining for Phase 2: the orchestrator persistent-spawn path (scope->workspace
clone, CMD = driver, registry.open on spawn, reap-on-confirm) — the deploy-side
piece, best finalized against a buildable image.
* feat(intake): orchestrator persistent spawn + start/stop for the live chat
Add the task-free spawn path for the intake (prompter) agent: one fixed
intake-1 container running the Agent-SDK driver (image ENTRYPOINT, not
claude -p), one live session at a time.
- spawn_intake_session clones the scope's repo(s) via WorkspaceService
(project -> one; product -> each distinct project, primary first),
composes the intake-1 prompt, resolves the model, and builds docker run
via _build_intake_run_cmd: no settings/hook mount (driver owns 9000),
no MCP config, no -w; registers the live relay and best-effort delivers
the opening message once the receiver is up.
- reap_intake_session closes the relay and stops the container.
- Routes: POST /live/start (project XOR product) and POST /live/{id}/stop.
- ROLE_MODEL_MAP[prompter]=opus; intake-1 -> roboco-agent-prompter image map.
- Replace the budget-sweep try/except/continue with _fetch_budget_status,
which logs the swallow at debug instead of silently dropping it.
25 new tests; docker + the clone are mocked. End-to-end container spawn is
pending a built image and the stack.
* feat(intake): wire /prompter to the live agent — scope form + SSE chat
Replace the Ollama chat loop on /prompter with the spawned-agent flow.
- IntakeForm: pick scope (project XOR product) + opening message + Start
before the chat; the agent clones that scope and reads the real code.
- use-prompter rewritten as the live brain (lib/api/prompter-live.ts): Start
spawns via POST /live/start, then an EventSource on /live/{id}/stream
streams the agent working — token deltas fill the assistant bubble,
tool_use/thinking drive a live activity line, a draft event renders the
existing DraftProposalCard. Messages go via POST /live/{id}/messages.
- Chat UX unchanged (Keep Chatting / Review & Confirm / ConfirmDialog reused);
reap-on-confirm and reap-on-leave call POST /live/{id}/stop.
- Drop the dead Ollama prompterApi client; trim prompter.ts to shared types.
Frontend gate green (tsc --noEmit, lint, build). The draft event + the
/live/{id}/confirm endpoint are the Phase 4 backend seam.
* feat(intake): confirm draft -> backlog task + agent draft emission
Complete the live intake vertical: the agent proposes a structured draft and
Review & Confirm turns it into a task.
- Draft emission: the prompter prompt instructs the agent to emit a fenced
roboco-draft JSON block when the spec is ready; the driver parses it into a
'draft' event over the existing relay -> the panel's DraftProposalCard. The
panel strips the raw block from the chat bubble.
- Fix a double-text bug: with include_partial_messages the reply arrives as
both StreamEvent deltas and the final AssistantMessage; the driver now takes
text from deltas only and the AssistantMessage for thinking/tool_use/draft.
- POST /live/{id}/confirm -> confirm_live_draft, reusing a draft->task core
extracted from confirm_draft; reaps the session on success.
- Both prompter confirm paths create at BACKLOG, not pending: backlog is the
holding area a draft waits in until it's reviewed and promoted to pending
(TaskService.activate). The legacy Ollama confirm was creating at pending,
skipping that gate — fixed.
- Remove the dead 'context' bootstrap param from the Ollama session-create
chain (schema + route + method + tests), superseded by the live scope form.
- No suppressions: replace every type:ignore/noqa across the intake surface
with a real fix (ORM .id -> UUID(str(x)); fakes -> monkeypatch.setattr;
lazy imports -> pyproject per-file ignore; union-attr -> recipients[0]).
Full make quality green; frontend tsc + lint green.
* build(intake): add the agent-prompter image builder to compose
The orchestrator references roboco-agent-prompter (AGENT_IMAGES + the
_ensure_agent_image dockerfile map) and docker/agent-prompter.Dockerfile
exists, but docker-compose.yml built every other agent image up front and
left this one out — so the image wasn't pre-built for a stack bring-up.
Mirror the other specialized agent-*-image builders: build from
docker/agent-prompter.Dockerfile, tag roboco-agent-prompter, depend on
agent-base-image.
* Created docker-compose.yaml for the NAS
* fix(intake): non-blocking /live/start so spawn never times out
The start POST awaited the whole spawn — workspace clone + first-time image
build + docker run — which blew past the panel's 60s HTTP timeout ('Request
timed out. The server may be busy.') and triggered a duplicate send. Found on
the 2026-06-09 NAS smoke.
- start_intake_session opens the live relay synchronously, then spawns the
container in the background (_spawn_intake_container_guarded). The route
returns the session id immediately; the panel opens the SSE stream right away.
- A background spawn failure is pushed onto the relay as an 'error' event and
closes the session, so the panel shows it instead of hanging.
- spawn_intake_session stays as the synchronous variant for direct callers/tests.
- Panel shows a 'Preparing the agent…' indicator until the first event arrives.
18 intake-spawn tests green; tsc + lint green. E2E re-validates on next smoke.
* fix(intake): propose_draft MCP tool + lock the agent down
Smoke 2026-06-09 exposed two compounding problems: the agent never reliably
emitted the draft (it narrated the spec instead of typing the magic fence), and
it had inherited the CEO's entire Claude Code env — Write/Edit/Bash + Gmail/
Notion/Calendar/Drive MCP — because bypassPermissions ignored the allowlist and
the mounted ~/.claude leaked the host MCP config.
- propose_draft: build_intake_options now registers an in-process SDK MCP tool
(create_sdk_mcp_server + @tool). The agent calls it to submit the draft; the
driver turns that ToolUseBlock into a 'draft' event (_is_propose_draft /
_draft_from_tool_input, tolerant of nested/flat/JSON-string input). The fenced
roboco-draft block stays as a fallback.
- Lockdown: strict_mcp_config=True + setting_sources=[] (ignore host MCP +
settings); permission_mode 'dontAsk' + a can_use_tool gate enforcing a hard
allowlist (Read/Grep/Glob/Task + propose_draft) replaces bypassPermissions.
- Prompt: call propose_draft (not a fence); the draft's downstream chain is
backlog -> Board (PO + HoM) -> CEO approve -> Main PM, and the agent's job ends
at the draft (it never routes or hands off).
SDK API verified against the installed claude-agent-sdk. Driver detection unit-
tested; the SDK-construction is validated on the next NAS smoke (incl. that
setting_sources=[] doesn't break the mounted-~/.claude auth).
* fix(intake): panel UX cluster from the smoke (#3/#4/#6/#12)
- #3 message boundaries: a tool call now ends the current text bubble, so the
agent's words before and after a tool render as separate messages instead of
one merged wall (the 'two waves merged into one bubble' the CEO saw).
- #4 activity indicator: promoted from tiny grey text to a prominent primary-
tinted pill so 'watch it work' is actually visible.
- #12 End chat: a header button (any chat state) reaps the agent and resets to
the form, reusing startAnother (which already stops the session). Backend
POST /live/{id}/stop already existed.
- #6 log noise: the opening-message delivery retry logs at debug, not error —
those failures are expected until the container receiver is up.
- Also fix a latent test gap from the #1 commit: the live-route test's fake
orchestrator now exposes start_intake_session (the route's non-blocking entry).
Frontend tsc + lint green; live-route + prompter_live tests green.
* fix(intake): render markdown in the chat bubbles (#8)
The agent emits rich markdown (### headers, **bold**, tables, lists) but the
bubble rendered raw text, so it was illegible (CEO-flagged on the smoke). Render
assistant content with react-markdown + remark-gfm (GFM tables) in a prose
container. Adds react-markdown + remark-gfm to the panel.
* feat(intake): #14 — two start routes (Board review vs straight to Main PM)
Per the CEO spec, the draft confirm now starts the task at PENDING with an
explicit assignment instead of parking it at backlog:
- route="board" (Board review & Start): assigned to the Product Owner, so the
orchestrator dispatches the full Board review (PO + Head of Marketing) before
the Main PM picks it up.
- route="main_pm" (Approve & Start): assigned straight to the Main PM, who
delegates to the cells (Board review skipped).
create_task_from_draft gains status + assigned_to params (default BACKLOG, so the
legacy confirm_draft is unchanged); confirm_live_draft + the /live/{id}/confirm
request carry the route. Service tests cover both routes.
* feat(intake): #14 draft-card buttons — Board review vs Approve & Start
Three buttons on the draft card now (CEO spec): Keep chatting / Board review &
Start / Approve & Start. The two action buttons confirm directly with their
route — launchTask(route) sends route to POST /live/{id}/confirm, which starts
the task at pending assigned to the Board (PO+HoM) or straight to the Main PM.
Supersedes the ConfirmDialog review step (scope is chosen up front in the form),
so it's removed from the page flow. The ConfirmDialog component + its sub-editors
are now unused — flagged for a follow-up cleanup, left in place to avoid churn.
tsc + lint green.
* fix(intake): keep the live SSE stream bound to its relay session
The orchestrator opened the relay session twice per live chat — once on the
request path (before the start call returns) and again inside the background
container spawn. The SSE stream binds to the session's queue the moment the
panel connects, so the second open swapped in a fresh queue and stranded the
stream: the agent replied normally, but its events went to the new queue while
the panel kept reading the old one, so the chat looked frozen on "Preparing…".
The second open was always redundant (the relay is opened by the caller before
the spawn). Remove it, and make open() idempotent so a live session is never
replaced out from under a stream that is already connected to it.
* fix(intake): draft-card launch buttons silently did nothing
The launch path required a `description` field, but the prompter draft schema
intentionally has none — it sends `objective` + the structured spec and the
backend composes the description (compose_description). `editableDraft.description`
was therefore undefined, so `description.trim()` inside launch validation threw a
TypeError that propagated out of the button's onClick. Clicking "Board review &
Start" / "Approve & Start" did nothing, with no feedback — the wall blocking the
whole confirm → task → reap flow.
- Map a proposed draft's description from `objective` as a fallback.
- Make launch validation null-safe.
- Replace the silent early-return with a toast that names what's missing, so a
blocked launch is never a dead, feedback-less button again.
* fix(intake): steer the agent to ask inline, not via AskUserQuestion
The intake's job is to ask clarifying questions, so it reached for the
AskUserQuestion tool — which isn't wired to the live chat panel and isn't in its
allowlist. The bare deny left it to stumble ("let me clarify… — no worries, let
me just lay it out") and waste a visible turn.
- Prompt: spell out that it asks by writing in the chat (the human reads every
message live) and that no question/prompt tool is available to it.
- Gate: give AskUserQuestion a specific deny message that nudges it to ask inline,
so even a reflex attempt degrades gracefully.
Also refresh the now-stale "what happens after propose_draft" section: the draft
card has three choices (Keep chatting / Board review & Start / Approve & Start)
and produces a pending task — not the old two-button "backlog" description.
* feat(intake): copy buttons on agent messages and the draft card
The CEO asked for a way to save the agent's plan/spec elsewhere "just in case" —
a cheap manual backstop until refresh-durability lands.
- New CopyButton: async Clipboard API when available, plus a legacy
textarea+execCommand fallback. The fallback is load-bearing — the panel is
served over plain http on a LAN IP, where navigator.clipboard is absent
(clipboard needs a secure context), so the modern API alone would never copy.
- Copy button under each assistant message (copies its text).
- Copy button on the draft card (copies the full spec as markdown: title,
objective, what-this-builds, the-work per cell, notes, success criteria).
* feat(intake): unbuffer logs + log each turn so the container isn't a black box
Debugging the intake smoke was painful for two reasons: (a) the orchestrator
block-buffered stdout, so `docker logs` lagged minutes behind reality, and (b)
the intake container logged only "session opened" then went silent for the whole
conversation (the chat streams to the relay, not stdout).
- Set PYTHONUNBUFFERED=1 on the orchestrator and agent-base images so structured
logs reach `docker logs` in real time instead of in large delayed chunks.
- Log each intake turn: "turn received" (with char count) and "turn streamed"
(chunk count + whether a draft was emitted), so the container logs show the
conversation's shape at a glance.
* chore(intake): remove the dead ConfirmDialog draft editor
The three-button draft card (Keep chatting / Board review & Start / Approve &
Start) replaced the old review-modal confirm flow, leaving ConfirmDialog and its
sub-editors (StringListEditor, TheWorkEditor) referenced by nothing but the
barrel export. Remove the three files and the export — typecheck + lint confirm
no remaining references.
* fix(intake): coerce bad draft enums on confirm instead of hard-failing
The intake agent is an LLM and will emit off-enum values — e.g. task_type="feature",
which is not a valid TaskType (code/documentation/research/planning/design/
administrative). `_coerce_draft_enums` called `TaskType(value)` directly, which
raised, and the confirm 400'd with "Draft has invalid or missing required fields:
'feature' is not a valid TaskType". That forced the agent to discover the valid
values and self-correct in-chat — unacceptable: clicking "Approve & Start" must
never blow up on a cosmetic enum guess.
Coerce each enum to a sane default on invalid/missing (task_type→code,
nature→technical, complexity→medium); team falls back to the first valid cell in
the_work, then backend. `_lead_cell_team` now skips invalid cell names too. The
confirm/launch action no longer hard-fails on an enum the model got wrong.
* fix(intake): draft card no longer renders above the user's latest message
attachDraft fell back to "the last assistant message anywhere" when the current
turn had no streamed text yet (propose_draft called first). That last message was
often the PREVIOUS turn's — sitting above the user's "Yes, propose it" — so the
draft card rendered above the user's message. Attach only to the current turn's
streaming message; otherwise append a fresh assistant message so the card always
lands at the bottom of the thread.
* test(intake): guard draft enum coercion + invalid-cell skipping
Regression tests for the confirm-time enum coercion: an off-enum task_type
("feature") / nature / complexity coerce to code/technical/medium instead of
raising, and _lead_cell_team skips invalid cell names. Locks in that a bad enum
guess from the agent can never 400 the launch again.
* fix(intake): stop the agent fumbling through Claude Code meta-tools
In smoke it reflexively probed CC built-ins before reaching propose_draft —
plan mode + ExitPlanMode (it announced a written plan and waited instead of
emitting the draft), ToolSearch, Write — each correctly denied by the lockdown
but stumbly, and it only proposed after explicit CEO nudges.
- Gate: ExitPlanMode now gets a specific deny nudge ("you don't use plan mode;
call propose_draft"), and the generic deny names the actual toolset instead
of a bare "not available", so any probe degrades into guidance.
- Prompt: forbid plan mode/ExitPlanMode/ToolSearch explicitly and spell out
"you do not plan and wait — call propose_draft directly when the spec is
ready," plus an anti-pattern bullet.
* feat(intake): make the container logs transparent mid-turn
`docker logs` on the intake container was a black box: only turn start/end, while
the agent read the codebase and spawned 20+ subagents invisibly (the conversation
streams to the relay, not stdout), and the benign 3x ~/.claude.json warning was
the only thing visible.
- Driver logs each tool call mid-turn ("Intake tool use" with the tool name) and
the draft emission, plus a tools count in the turn-streamed summary. Text deltas
stay unlogged (they'd spam). Now the logs show the turn's real shape.
- Pre-create ~/.claude.json ({}) at container boot so the CLI's "config not found"
warning (printed 3x, self-healed anyway) stops drowning the real logs.
* fix(intake): render markdown in user messages + scope copy to code blocks
Two display fixes from the smoke:
- User messages collapsed newlines (plain {content} in a div) and rendered no
markdown — a "1.\n2.\n3." answer showed as one run-on line. Render user AND
assistant bubbles through a shared GFM markdown body that inherits the bubble's
text color, so lists / newlines / styling render correctly on both.
- Copy was blanketed on every assistant message; scope it to KEY parts — a copy
button on fenced code blocks (the draft card keeps its own). Removed the
per-message button.
* fix(intake): prevent duplicate tasks from a double-click on launch
Clicking a draft launch button twice fired two confirms and created duplicate
tasks. Add a synchronous re-entry guard (a ref — no stale-closure window) at the
top of launchTask so a second click returns immediately, and disable + spin the
draft-card buttons while a launch is in flight so it's visually clear it's working.
* docs(how-to): lead task creation with the Task Assistant flow
Rewrite "1 · It starts with you" to walk the Prompter/Task Assistant path —
scope form, the agent reading the codebase, its grounded analysis, the draft
card, and the created task — then flow into the Board review. Replaces the old
manual task-definition form shots.
Image placeholder: images/prompter_draft_card.png (the 3-button card) is
referenced but not yet captured — TODO comment marks it for the next smoke run.
A second comment flags an optional re-capture of prompter_run_2 after the
markdown-rendering fix.
* fix(intake): restore assistant message text contrast
The markdown refactor dropped `dark:prose-invert` and made text inherit the
bubble's color, but the assistant bubble had no explicit text color — so its text
rendered near-invisible (dark-on-dark on bg-muted). Give the assistant bubble an
explicit text-foreground; the user bubble already carries text-primary-foreground,
and [&_*]:!text-inherit now resolves to a readable color on both.
* fix(intake): coerce draft priority too — confirm 500'd on priority="high"
The enum-coercion fix covered task_type/nature/complexity/team, but priority is a
non-enum int field handled by `int(draft_data.get("priority", 2))`, and the agent
guesses a word ("high") as readily as a number — so int("high") raised ValueError
and the confirm 500'd. Same class of bug, one field missed.
Add _coerce_priority: map words (urgent/high/medium/low → 0/1/2/3), clamp numbers
to 0-3, default to 2 (medium) on anything else. The launch can no longer crash on
any field the LLM guessed. + regression test.
* fix(intake): draft card shows distinct cells, not one badge per work item
the_work has one entry per work item, so a cell with several items rendered its
badge repeatedly ("Board-led across Backend Backend Backend Frontend Frontend
…"). De-dupe to distinct teams so the card reads "Board-led across Backend
Frontend" — and the "Cell:" vs "Board-led across" label keys off distinct count.
* docs(how-to): hero the teaser gif + resolve the Prompter/Task Assistant thread
- Move the 12s teaser gif to the top as the hero — it was buried between the
"prefer video" link and the first screenshot.
- Name the connection: the Task Assistant IS the Prompter, so section 1 (using
the tool) and the rest (RoboCo building it) read as one story — you use the
tool the company built for itself, then watch the build.
- Re-anchor the section 1 → Board transition to follow the Prompter's own
journey, instead of implying section 1's example task is the one reviewed next.
* Included images for how-to.md
* docs(how-to): align agent count to 20 (matches README + CLAUDE.md)
The how-to said "18 agents" with UX/UI at one dev and no Intake — stale against
the authoritative count. Bump 18→20 (prose + spelled-out eighteen→twenty), give
UX/UI 2 devs, and add the Intake line to the org tree (Intake leads section 1, so
it belongs in the tree). README + CLAUDE.md already say 20.
* ci(release): publish all RoboCo images to GHCR + Docker Hub
The release published only the orchestrator to GHCR. Build and push the full set
the stack needs — agent-base, the 8 agent images, orchestrator, and panel — to
BOTH ghcr.io/rennf93/* and docker.io/renzof93/*, at :<version> and :latest, so
consumers can pull instead of compose-building.
- agent-base builds first (the agent images build FROM roboco-agent-base, a local
tag), then the rest; push only after every build succeeds.
- Image names mirror the docker-compose `image:` values 1:1.
- Free disk on the runner first (11 images is space-heavy).
- Needs a DOCKERHUB_TOKEN repo secret for the Docker Hub login.
- SECURITY.md updated to reference both registries.
* ci(release): use short SHA as the image tag on manual dispatch
A workflow_dispatch runs against a branch, and the branch name (e.g.
feature/prompter-gold-upgrade) was used verbatim as the image tag — but "/" is
illegal in a Docker tag, so the first build failed instantly with "invalid
reference format". Releases still tag from the release tag; manual dispatch now
always uses the short SHA, which is a valid tag.
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
|
3205443119 |
Fix: dependency spawn gate and cell ownership (#73)
* Cleanup + Missing greenlet error * fix(messaging): persist a group's active-session pointer so posts reuse it create_session and create_session_with_access_check set group.active_session_id from session.id BEFORE the flush that materializes it — the id is a flush-time uuid4 default, so the pointer was written as NULL and every post opened a fresh session, fragmenting one conversation across many. Flush first, then link, the same ordering the seed path already uses. Two tests fabricated "two distinct sessions" by calling create_session twice on one group, which only differed because of this bug; switch them to two groups so they keep testing their real intent. Add a regression guard that the pointer is actually persisted and a second create reuses the live session. * fix(orchestrator): gate spawns on dependencies and keep cell tasks in their cell The cross-task dependency check ran only on the dev dispatch path, so cell-PM, Main-PM and board agents were spawned onto dependency-blocked tasks and flailed unblock / escalate / notify against an unfinished upstream — climbing ownership of cell work up to the board, which cannot drive it, and deadlocking the task. - Move the dependency gate into the shared spawn readiness check so it covers every role, and auto-block the task so it leaves the pending pool until the upstream reaches a terminal state (then the existing auto-unblock revives it). - Cell-ownership invariant: a backend/frontend/ux_ui task may only be worked or owned by its own cell. The readiness gate refuses a board or Main-PM spawn onto a cell task; reassign refuses and clears such an owner; and on dependency-clear a mis-owned cell task is re-homed to its cell's pending pool instead of reviving under an owner that cannot progress it. - A dependency block is never a CEO signal: notify(target=ceo) is refused while the task is waiting on an unfinished upstream, with a remediate to idle and wait — the block clears on its own. * Uploading images + Fixing pyproject.toml * ++ * revert(orchestrator): drop the cell-ownership block pending a tooling audit The cell-ownership invariant added earlier — a board / Main-PM role may never be spawned onto or reassigned to a cell task, plus re-homing a mis-owned cell task on dependency-clear — was too absolute. It forbids a higher role from stepping in when something genuinely deeper is going on, and contradicts the existing rule that main_pm may hold a task at awaiting_pm_review. The dependency spawn gate already prevents the cascade that handed the board cell tasks; the deadlock it guarded against will be addressed with a return-path approach after auditing what tools the cell PMs actually need. Keeps the dependency gate and the CEO dependency-block notify guard. * docs(prompts): a dependency wait is wait-and-idle, not escalate The cell-PM and Main-PM prompts told agents to escalate_up / retry unblock on a blocked task without distinguishing a dependency wait (which auto-clears the moment the upstream completes) from a real wedge — the source of the escalate/unblock flail and the CEO-notification spam. Split the blocked-state guidance: a cross-cell dependency wait = note + i_am_idle (do not escalate, unblock, or notify the CEO); escalate only a genuinely broken upstream. Fix two stale references to i_am_blocked, a developer-only verb the PMs do not have, to escalate_up. Correct the CLAUDE.md verb-surface table, which understated every role: it listed 4 cell_pm verbs while the flow manifest derives the full set (11, including unclaim and i_am_idle) from lifecycle.spec.intents_for_role. * feat(gateway): cell_pm reassign verb — intra-cell developer hand-off A cell PM can now hand a claimed/in_progress task to another developer in its own cell without unclaim (which drops the work back to the pool and loses the assignee). The branch is keyed to the task, so the work-in-progress is preserved; the new dev is respawned to continue. Intra-cell only: the task must be in the caller's cell and new_assignee must be a developer of that same cell. Wired through every layer: the reassign IntentSpec (composes=(), cell_pm-only), the choreographer verb + intra-cell guard, a reaper-safe TaskService.reassign_active_claim (reseeds the claim heartbeat so the new dev is not immediately reaped), the ReassignRequest schema, the cell_pm flow route, and the MCP flow-server tool. Tracing-waived like unclaim (mechanical hand-off). Regenerated lifecycle/verb artifacts; prompt + CLAUDE.md updated. --------- Co-authored-by: Renn F <rennf93@users.noreply.github.com> |
||
|
Renn F
|
073c2ec8b5 |
fix(prompts): repair the verb-table generator and regenerate
regenerate_verb_tables.py imported roboco.api.schemas.v2, which no longer exists (schemas moved to v1), so it raised on import and the generated verb/tool tables could never be refreshed — leaving _generated/verbs.md and the per-role prompts stale (e.g. listing submit_for_qa, omitting the notify_*/channels/progress/pr_update content tools). Repoint the imports to v1, fix the renamed schema (OpenPrRequest), and regenerate. |
||
|
Renn F
|
fd4df51572 |
docs: correct doc-vs-code drift across the canonical docs
A documentation audit against the code surfaced several stale claims: - Agent count: the roster is 19 AI agents (the UX/UI cell has two devs, ux-dev-1 + ux-dev-2), not 18 / a single UX dev. Fixed in README, CLAUDE.md, base.md, and docs/ux_ui. - API: domain routes are mounted under /api, not /api/v1 (the /api/v1 prefix is the agent gateway only); dropped the non-existent /api/v1/test group; fixed the orchestrator-status path in deployment.md. - Quick Start uvicorn target is roboco.api.app:app (the api package deliberately does not export app). - Verb table: the developer PR verb is open_pr (renamed from submit_for_qa); the lifecycle's canonical module is foundation/policy/lifecycle.py (enforcement/task_lifecycle.py is a shim). - Backend team stack: vector store is PostgreSQL + pgvector (via piragi), not Qdrant; mypy targets roboco/, not src/. - .env.example: replaced the phantom Qdrant/OpenAI blocks with the real Ollama/RAG settings. |
||
|
Renn F
|
4c59bfa840 |
fix(prompt): Main PM delegate description is a goal+constraints brief, not a solution-dump
delegate already requires substantive acceptance_criteria (TASK_AT_CREATE completeness), so the defect was not missing criteria — it was the Main PM writing the cell-PM subtask description as prescriptive prose that dictates the cell's solution (e.g. a full UX layout for a design task), doing the cell's job and wasting the expertise it delegated to. The role doc now tells the Main PM the description is a brief: state the goal + the constraints to fit, leave the HOW to the cell, and for design tasks give the problem, not a mockup. |
||
|
|
110aaa7a77 |
Chore: v1 removal gateway canonical (#46)
* chore(agent_sdk): remove dead /traceability/remind endpoint and reminder map
The TRACEABILITY_REMINDERS dict and its /traceability/remind endpoint were
keyed entirely on pre-gateway tool names (roboco_task_*, roboco_journal_*,
roboco_message_send, roboco_session_create_for_tasks) deleted in the gateway
cutover. The endpoint had zero callers; v2 enforces traceability server-side
in the Choreographer.
* fix(bootstrap,seeds): onboarding prompts call give_me_work(), not deleted roboco_task_scan()
The startup prompt and the seeded cell/all-hands channel onboarding messages
instructed agents to call roboco_task_scan() — a tool removed in the gateway
cutover. Point them at the live give_me_work() flow verb.
* fix: replace remaining deleted v1 tool names with gateway verbs
Spawn prompts, onboarding strings, remediation messages, and comments still
referenced pre-gateway tools deleted in the cutover (roboco_task_*,
roboco_agent_idle, roboco_notify_*, roboco_message_send,
roboco_session_create_for_tasks, roboco_journal_*, roboco_escalate). Rewrote
each to the correct role-scoped gateway verb (give_me_work/i_will_work_on for
workers, triage for PMs, i_am_done vs complete, notify/notify_ack, escalate_up,
unclaim, i_documented, open_session, note). Updated one enforcement-message
test that matched the old tool name by coincidence.
* test: guard against deleted v1 tool names reappearing in roboco/
Scans roboco/ for the deleted pre-gateway tool names; excludes the orphaned
roboco/agents/ subtree (removed in a later phase).
* chore(exceptions): drop 8 unused pre-gateway exception classes + their tests
LLMError, RAGError, AlreadyExistsError, TaskBlockedError, TaskClaimError,
AgentNotAvailableError, AgentBusyError, NotificationPermissionError were never
raised in production. SessionClosedError/DatabaseError are kept (live + tested).
* chore(models): drop unused pre-gateway notification/channel/handoff factories
Removes create_task_assignment/_blocker_escalation/_review_request/
_documentation_request/_priority_change/_alert/_broadcast, create_cell_channel/
_cross_cell_channel/_announcements_channel, create_handoff (+ HandoffParams),
ProactiveContext, and A2APartType. The gateway choreographer builds these
server-side now. Drops the matching dead-code tests.
* chore(services): drop unused pre-gateway permission/messaging/audit/optimal/remediation methods
These pre-gateway helpers (channel-permission checks, channel-membership ops,
permission-denial audit hooks, doc ingestion, two remediation hints) have no
production caller — the gateway role_config + enforcement layer replaced them.
Drops the matching dead-code tests; live methods (send_message, the SESSION_*
flow, log_task_action_denial, etc.) are untouched.
* chore(orchestrator,ws,events,config): drop unused pre-gateway lifecycle/broadcast/roster symbols
orchestrator: get_running_agents, is_agent_busy, queue_priority_work,
get_all_instances (+ their OrchestratorAccessProtocol declarations in events.py).
websocket: broadcast_new_message, broadcast_session_closed (no event type emits
them). agents_config: ALL_PMS/ALL_DEVS/ALL_QA/CELL_PMS roster constants (ALL_DOCS
stays — it gates docs-write workspace perms).
* refactor(agents): delete orphaned pre-gateway agent subtree + dead organization model
The Gateway/full cutover replaced the Python agent-class implementations with
the server-side Choreographer; the classes survived only as a self-referential
island. Removes roboco/agents/{base,mixins,factory,board,developer,documenter,
pm,qa,orchestrator}.py and roboco/agents/factories/{board,cells,developers,
documenters,pms,qa}.py, plus roboco/models/organization.py (Cell/Board/
Organization — used only by those factories). Keeps factories/_base.py
(compose_prompt — the live prompt-layer composer the orchestrator calls at
spawn) behind minimal package __init__ files.
* chore(db): drop dead tasks.execution_log + outputs columns (migration 015)
Both JSON columns had zero readers/writers in code, tests, and migrations —
execution progress is tracked via progress_updates and artifacts via
commits/documents. Removes the ORM columns, the Pydantic Task.execution_log/
outputs fields, the ExecutionLog/FileRef models (+ their __init__ exports), and
the now-invalid kwargs from test fixtures. Migration 015 (down_revision
014_drop_pm_approvals) verified live: upgrade drops, downgrade re-adds.
Apply on the NAS with 'alembic upgrade head' at next deploy.
* chore(config): drop 16 unread Settings fields
Verified unused (no settings.X, no self.X property use, no getattr-by-name):
app_name, reload, workers, openai_api_key, secret_key, access_token_expire_minutes,
algorithm, log_level, log_format, the four session_* limits, message_max_length,
commit_subject_min_chars, commit_banned_words, agent_budget_sweep_interval_seconds.
Removes the empty Logging + Sessions&Messages sections and orphaned .env.example
vars. Kept: redis_db/redis_password (redis_url property), agent_sla_* (read via
getattr in task_lifecycle), encryption_key, and all live thresholds.
NOTE: commit_banned_words/commit_subject_min_chars and
agent_budget_sweep_interval_seconds were feature-config never wired to their
consumer (commit validator / budget sweep) — removed as dead, but flagged in
case the intent was to wire them.
* test(lifecycle): give i_will_work_on calls a substantive plan (#171 contract)
The real-DB lifecycle tests called i_will_work_on with a 13-char plan and no
risks/technical_considerations, so the substantive-plan gate (#171) rejected
them with incomplete_input — failing on master. Supply a >=150-char plan plus
technical_considerations and risks (mirroring tests/unit/gateway/
test_choreographer_dev.py). All 6 now pass; gate runs with no deselect.
* feat(gateway): wire commit-validator thresholds to settings
commit_subject_min_chars and commit_banned_words were config defined but never
read — the gateway commit() gate used the validator's hardcoded module defaults.
Re-add the two Settings fields and pass them through validate_commit_message in
content_actions.commit(), so config is the source of truth (validator defaults
remain the standalone/CI fallback). Adds wiring tests that monkeypatch settings
and assert the gate honors them.
* refactor(orchestrator): retire gateway_enabled flag; trigger_filter is unconditional
The gateway_enabled Settings field gated only the trigger_filter spawn-cooldown
(never the agent tool surface). Prod ran it on; the Phase-0 'legacy dispatch
path' it guarded no longer exists. Remove the field + the early-return branch in
gateway_pre_spawn_check so the cooldown runs for every spawn, drop the now-dead
ROBOCO_GATEWAY_ENABLED from docker-compose.yml, and update the stale Phase-0
comments + cooldown test. The per-container ROBOCO_GATEWAY_ENABLED env (set by
_append_manifest_args, read by agent_sdk to load the manifest) is unaffected.
* refactor(api): relabel /api/v2 -> /api/v1 as the canonical gateway surface
The gateway is the only agent API now, so the 'v2' label (with no v1) was
misleading. Renames roboco/api/routes/v2 -> routes/v1, schemas/v2 -> schemas/v1
(+ the matching test dirs and test_v2_role_dep/test_schemas_v2_flow files),
rewrites every /api/v2 path, routes.v2/schemas.v2 import, and v2-* router tag to
v1, and refreshes the stale 'v2' comments/docstrings. The panel is untouched (it
uses the unversioned /api/* REST routes). flow_server/do_server now POST to
/api/v1/*.
* docs(scripts): reset_runtime_state header matches actual SQL behavior
The header claimed it preserves groups + journals, but the .sql wipes both
(verified live: groups 6->0, journals 5->0; only agents/projects/channels
survive). Correct the wiped/preserved lists to match.
* refactor(gateway): extract _build_rich_plan to drop i_will_work_on under the complexity gate
i_will_work_on was cyclomatic rank C (11) — one over the xenon --max-absolute B
threshold — because of the five `x or default` fallbacks in the rich_plan dict.
Move that dict into a small _build_rich_plan helper (behaviour identical); both
methods are now rank B. make quality is fully green (xenon was its last failure;
bandit already passed — its 34 findings are all LOW severity, filtered by -ll).
* feat(foundation): add canonical CELL_TEAMS set; dedupe cell-subset literals
* feat(db): add ProductTable + ProductProjectTable ORM (per-cell project map)
* feat(task): add additive nullable product_id (ORM + model + DTO + create threading)
* feat(task): thread product_id through create_subtask/route/response
* feat(db): migration 016 — products, product_projects, tasks.product_id
* fix(db): document migration 016 plan deviations (revision len, FK name)
Two values in migration 016 intentionally diverge from the Task 2.4 plan
literals; this strengthens the in-file justification so the deviations are
self-documenting and verifiable.
- revision id (plan line 623): the plan's 36-char
"016_add_products_and_task_product_id" overflows alembic_version.version_num
(VARCHAR(32)) — alembic upgrade head raises asyncpg
StringDataRightTruncationError. Kept at 27 chars
("016_add_products_product_id") so Step 4's live round-trip stays green.
- downgrade FK name (plan line 683): roboco/db/base.py sets a metadata
naming_convention, so the FK upgrade() creates is
"fk_tasks_product_id_products", not the Postgres default
"tasks_product_id_fkey". The plan literal does not exist in the DB and
would fail the downgrade with "constraint does not exist".
Both verified via the live upgrade/downgrade round-trip on a throwaway DB.
Issue 3 note: the prior commit (b896cac) also touched
tests/unit/api/test_schemas_tasks.py (added product_id=None to the
task_to_response stub). That line is load-bearing — task_to_response reads
task.product_id (added in Task 2.3, commit 67afa6b) — and belongs to Task 2.3's
scope; it is left in place because removing it breaks 4 tests and history is
not rewritten.
* refactor(db): trim migration 016 deviation notes to plan-faithful form
Reverts the out-of-scope documentation expansion (commit 1a4f296), which
was a second undocumented commit beyond Task 2.4's single plan-specified
commit and only bloated the migration docstring/comments.
The migration file now matches the plan-specified commit (b896cac) byte for
byte: the two necessary deviations from the plan literals stay (revision id
shortened to fit alembic_version.version_num VARCHAR(32); downgrade FK name
follows db/base.py's metadata naming_convention), each kept to a concise
inline note in the plan's header style.
The Task 2.3-scoped test stub line (tests/unit/api/test_schemas_tasks.py
product_id=None) is load-bearing — task_to_response reads task.product_id —
and is left in place; history is not rewritten.
Verified: live alembic upgrade head + downgrade to 015 round-trip on a
throwaway DB drops products/product_projects/tasks.product_id cleanly, and
make quality is green.
* refactor(test): annotate db_session and drop type: ignore in migration 016 test
Annotate the test_products_tables_and_task_fk_exist param as
db_session: AsyncSession (imported under TYPE_CHECKING) and remove the
# type: ignore[no-untyped-def] suppression, matching the typed db_session
pattern used across tests/integration/.
* feat(models): Product + ProductCreate/Update + ProductCellMapping (cell-validated)
* refactor(models): minimize ProductCellMapping config override to use_enum_values
The previous override re-declared validate_assignment, populate_by_name,
and extra=forbid, which RobocoBase already supplies. Pydantic merges
model_config across inheritance, so overriding only use_enum_values=False
is sufficient to keep team as a real Team enum (required so team in
CELL_TEAMS and enum identity hold for callers) while inheriting the rest
of the base config.
* fix(models): document ProductCellMapping use_enum_values override as plan-mandated
Resolves SPEC-COMPLIANCE review notes for Task 3.1 (Product domain models).
1. The ProductCellMapping use_enum_values=False override is a deviation from a
bare project.py mirror, but it is mandated by the plan's own Task 3.1 code:
RobocoBase sets use_enum_values=True, which coerces team to the plain string
"backend". The plan's Step 1 test asserts m.team is Team.BACKEND (enum
identity) and the Step 3 validator formats its error with v.value, both of
which require team to remain a real Team enum. The override is therefore
necessary; this commit relabels the comment to cite the specific spec lines
that force it instead of leaving it as an unexplained departure. Downstream
Task 3.2 (_replace_cells / project_for) already tolerates either form and the
ORM stores the same value regardless, so the override has no behavioral reach
beyond the in-memory enum identity the plan's test checks.
2. test_product_model.py hoists 'from uuid import uuid4' to module level rather
than inline (as the plan's verbatim Step 1 code shows) because the global
Pylint PLC0415 rule (import-outside-top-level) forbids inline imports and
there is no per-file-ignore for tests/unit/models/. The hoisted form is the
only ruff-clean rendering of the plan's test; left unchanged here.
3. Task 3.1 landed across two commits (c616d95 create, 6ebad255 refactor) rather
than the plan's single Step 5 commit. Earlier history is intentionally not
rewritten; this single follow-up commit brings the model to its final
spec-faithful, fully-documented state.
* feat(service): ProductService CRUD + project_for per-cell resolver
* feat(api): Product CRUD routes + schemas, wired into the app
* fix(api): roll back and map cell-replacement IntegrityError on product update
update_product replaced cells via ProductService._replace_cells without
any try/except, so a duplicate-team cell (uq_product_projects_product_team)
or a non-existent project_id (product_projects.project_id FK) raised an
IntegrityError at flush, poisoning the AsyncSession and surfacing an
unhandled 500 with no rollback. Wrap the update + commit in a try/except
that rolls back and maps the UNIQUE violation to 409 and the FK violation
to 422, mirroring create_product's rollback discipline. Add integration
tests covering both client-error paths.
* fix(api): map create_product cell-mapping IntegrityError to 409/422
create_product only caught the slug conflict ('already exists' in str(e))
and bare-raised everything else, so a cells entry whose project_id does not
reference any project let the product_projects.project_id FK IntegrityError
propagate out of the route as an unhandled 500. The matching update_product
path was already hardened (uq_product_projects_product_team -> 409, FK
violation -> 422); apply the same mapping in create_product so a bad
project_id (or a duplicate-team cell) is a client error, not a server error.
The slug conflict is now caught as ConflictError directly instead of via a
broad except + string match.
* feat(gateway): add optional project_id to delegate inputs/request/routes
* feat(gateway): per-cell project routing (override -> product map -> parent) + product_id inheritance
* feat(task): approve_and_start — reassign board task to Main PM (CEO gate #1)
* feat(api): POST /tasks/{id}/approve-and-start (CEO gate #1, notes-required)
* test(api): cover approve-and-start 404-before-notes-gate for missing task
* feat(panel): Product types + Task.product_id
* feat(panel): productsApi + hooks + tasksApi.approveAndStart
* feat(panel): Products management screen + sidebar nav
* feat(panel): Approve & Start button (CEO gate #1)
* fix(api): narrow delete_product to IntegrityError + cover 204/409 delete paths
* test(task): assert approve_and_start persists + appends the audit note
* refactor(db): migration 016 names the tasks.product_id FK explicitly (house style)
* fix(db): make migrations authoritative + self-heal orphan product tables
init_db() no longer silently falls back to create_all when alembic upgrade
fails. That fallback masked migration failures and, since create_all cannot
ALTER an existing table, left the schema inconsistent — turning an unapplied
migration 016 into a crash loop: 016's CREATE TABLE products failed, the
upgrade rolled back, create_all re-created an empty orphan products table, and
every later boot failed again on the now-existing table while tasks.product_id
never got added. Now a migration failure is raised so the real error surfaces.
Migration 016 additionally drops EMPTY orphan products/product_projects tables
left by the old fallback before creating them, so an already-polluted DB
self-heals on the next deploy with no manual SQL. Skipped in offline (--sql)
mode; refuses to drop a table that holds rows.
* fix(db): create_all is the schema source of truth; alembic for increments
The Alembic chain is incomplete relative to the ORM — columns/tables like
notifications.delivered_at and the RAG indexed_documents table have NO migration
and have only ever been materialized by create_all. Tests don't catch this
because the test DB is also built via create_all, so migrations are never
exercised. The prior 'migrations are authoritative' init_db (and before it, the
create_all-only-on-failure fallback) therefore left a migrate-only boot with
missing columns/tables.
init_db now reflects reality:
- Fresh DB -> create_all builds the full current ORM schema, then stamp
Alembic at head so later incremental migrations apply.
- Existing -> run pending migrations (a real failure is raised, not masked),
then create_all(checkfirst) to gap-fill any missing ORM tables.
create_all cannot add a column to an existing table, so an ORM column added
without a migration needs a fresh rebuild of that table to appear.
* fix(db): migration 017 reconciles the Alembic chain with the full ORM schema
For years the live schema was built by create_all, not migrations, so the chain
drifted — tables/columns/indexes in the ORM had no migration (the
indexed_documents table, notifications.delivered_at, ~15 indexes, plus
timestamptz/server-default metadata). With init_db no longer masking that via a
create_all fallback, a migrate-only boot was missing those objects.
017 was produced by 'alembic revision --autogenerate' against Base.metadata,
reviewed, and verified: on a fresh DB, 'alembic upgrade head' (001..017) now
reproduces the create_all schema EXACTLY — a re-run of autogenerate detects zero
changes — and the 017 upgrade/downgrade round-trips cleanly. The migration chain
is now complete: migrate-only and create_all converge.
Also updates the init_db tests to assert the new behaviour (raise on an existing
DB's migration failure; create_all + stamp head on a fresh DB) instead of the
removed silent fallback.
* feat(panel): Product picker in the New Task form (drives per-cell routing)
The Products screen and Approve & Start button shipped, but the task-creation
form had no way to attach a Product — so a human couldn't set product_id from
the UI, which is exactly what drives per-cell project routing of delegated
subtasks. Adds an optional Product dropdown (Advanced -> Git config) populated
from useProducts(); 'None' falls back to the single project.
* fix(db): seed data is preserved on a fresh DB (run migrations, not bare create_all)
The previous fresh-DB path (create_all + stamp head) built the tables but never
ran the migration chain, so migration-embedded SEED DATA was skipped — most
visibly the AI providers seeded in 004. After a DB reset that left
provider_configs empty, so PUT /api/providers/ollama-key 404'd (the handler
raises NotFoundError when the Ollama provider row is missing).
Since migration 017 made the chain reproduce the full ORM schema, init_db now
runs 'alembic upgrade head' from base on a fresh DB — building every
table/column/index AND running the seeds. Verified: a fresh upgrade head seeds
both provider rows. Existing DBs still get migrations + create_all gap-fill.
Updates the init_db fresh-DB test accordingly.
* feat(task): project_id optional when a product_id is set (board fan-out tasks)
A board task that fans out across cells via a Product has no single repo of its
own — backend/frontend/ux_ui are each wrong, because the root coordinates and
delegates. Forcing one arbitrary Project was broken design (flagged at design
time). project_id is now nullable; a task must have project_id OR product_id:
- TaskCreate model validator + a TaskService.create() invariant (covers every
create path).
- ORM/DTO/schema: project_id nullable; task_to_response uses to_python_uuid.
- Gateway: a parent with only a product can delegate (guard now needs BOTH
project and product to be None to reject); _resolve_subtask_project resolves
each subtask from the product map and raises a clear error if a cell has no
mapping and no parent project.
- Migration 018 (tasks.project_id nullable), round-trip verified; fresh
upgrade head still seeds providers.
- Panel: Project no longer required once a Product is selected.
- Removed the dead, never-called a2a create_task_from_message (it could only
ever create a repo-less task) + its two coverage-only tests.
make quality green; panel tsc/lint/build green.
* Upgrade to Minimax M3
* fix(db): seed providers on existing DBs + correct enum casing
Migration 004 created the modelprovider/assignmentscope enums and seeded
provider rows in UPPERCASE, but the ORM (_str_enum) reads/writes the
lowercase StrEnum .value — so a fresh migrate-from-base DB built an enum
the ORM cannot read. Lowercase the enum labels and seed values in 004.
Add idempotent migration 019 to (re)seed the Anthropic + Ollama Cloud
providers with ON CONFLICT (name) DO NOTHING, so an existing DB whose
provider_configs table was created by create_all (and never ran 004's
seed) gets the rows on the next `alembic upgrade head` — fixing the
/api/providers/ollama-key 404 without a volume wipe.
* fix(tasks): let board/fan-out coordination tasks flow without a repo
A coordination task (project_id NULL, product_id set) targets no repo of
its own — it fans out to cell subtasks that each resolve a real project
from the product's cell->project map. Several paths still assumed every
task does git work and blocked it:
- orchestrator: add _is_coordination_task() and exempt these tasks from
the project/branch/git-token gates in _readiness_check_task,
_readiness_gate, _check_stuck_conditions, _validate_task_for_spawn.
- services/task.py: _ensure_branch_for_task returns "" (no branch) for a
coordination task instead of raising; activate requires project OR
product. This unblocks Main PM's i_will_plan claim, which otherwise
raised before it could delegate the fan-out.
- gateway: _pending_assignment_guard exempts advisory roles
(product_owner/head_marketing/auditor) from the "assigned but never
claimed" idle gate — they review without claiming, so they could not
satisfy a claim-or-unclaim remediation.
Adds focused unit tests for each.
* fix(tasks): coordination tasks reach in_progress + team reflects Main PM
The board->cells fan-out deadlocked: a coordination/fan-out task (product set,
no project of its own) could be created and claimed, but start()'s
claimed->in_progress transition hit validate_git_requirements, which still
demanded a branch_name and raised GitRequirementError. So Main PM's i_will_plan
never completed — it looped and never delegated. c961282 exempted
_ensure_branch_for_task (branch creation) but missed this parallel git gate in
the enforcement layer.
- task_lifecycle.py: add GitContext.is_coordination; skip the
claimed->in_progress branch_name gate when it is set.
- task.py: populate is_coordination=(project_id is None and product_id is not
None) in _validate_and_set_status; a branchless code task is still gated.
- approve_and_start: set team=Team.MAIN_PM on hand-off so the task isn't left
labelled team=board after it leaves the board (now assigned to main-pm).
Adds a lifecycle-gate unit test and an end-to-end integration test that
claims, plans, and starts a project-less coordination task.
* fix(hooks): remove dead traceability hook + stale deleted-verb references
The v1-removal cleanup (2cfbf39) deleted the /traceability/remind SDK endpoint
but left the PostToolUse hook that curls it, so every gateway tool call 404'd
and agents silently lost their traceability reminders. Remove the dangling hook
(registration + TRACEABILITY_TRIGGER_TOOLS + Dockerfile COPY + the script); v2
carries per-verb guidance on the Envelope. Also correct two stale pre-gateway
tool names in hook text: the budget loop-detector nudged agents toward the
deleted roboco_task_escalate() (now unclaim()/i_am_idle(), which every looping
role has), and an sdk-startup comment referenced roboco_task_scan/get.
Extends the deleted-tool-name guard to scan docker/scripts/*.sh and to assert
every $SDK_URL/<path> a hook curls is a route still served by the SDK — the
check that would have caught this class (it lives in shell, invisible to mypy
and the Python import graph).
* fix(db): backfill ORM enum values the migration chain never added
Several StrEnum values were added to the ORM over time without a matching
`ALTER TYPE ... ADD VALUE` migration; 017 was autogenerate-derived and
autogenerate does not detect added enum labels, so the drift survived. On a DB
whose enum type predates the value, binding it raises at runtime — e.g.
`invalid input value for enum notificationtype: "a2a_request"` on
GET /api/notifications (list_system_notifications), and the same class for
blockerresolvertype/handoffstatus/team.
Migration 020 adds every drifted value idempotently (ADD VALUE IF NOT EXISTS —
no-op when 009 already reconciled it). Runs on the next `alembic upgrade head`.
Detected by comparing each ORM enum's values to the labels the migration chain
produces; adds tests/unit/test_enum_migration_parity.py which renders the chain
offline and fails on any future drift — the check that would have caught both
this and the provider-enum bug.
* fix(orchestrator): stop branch auto-block, board reassign, unblock livelock, agentless claims
Cluster C1 — four coupled orchestrator/task-invariant defects:
#18: a branch is created only at claim, so a pending, never-claimed code task
legitimately has no branch_name. The stuck-detection sweep (pending-only) and
readiness gate flagged that as "Task missing branch_name" and auto-blocked the
task every tick, so it never dispatched. Centralize the gate in
_branch_is_expected (status in claimed/in_progress/verifying, never a
coordination task) and apply it in both _check_stuck_conditions and
_readiness_check_task.
#14: the main_pm -> product_owner escalation rung handed an in_progress
descendant code task to the Product Owner (a board role) and marked it BLOCKED;
the board has no verb to own code work, so the dev's finished work deadlocked.
TaskService.apply_escalation (the single write primitive — covers both the
gateway escalate verb and the HTTP escalate route) now diverts a descendant code
task targeting a board/advisory role: it releases the task to PENDING for a
role-matched cell claim instead of stranding it.
#17: a blocked task reassigned to Main PM kept respawning the ex-assignee cell
PM to unblock it, but the assignee-only pre-unblock note returned not_authorized
— a livelock. _dispatch_blocker_work now dispatches the task's CURRENT PM/board
assignee (the unblock authority), falling back to the cell PM only when no
PM/board holds it. Also: a branchless coordination parent yields no valid merge
target — resolve_parent_branch now falls back to the child's own project default
branch (e.g. master) via TaskService.project_default_branch_for_task, and
_check_parent_branch_ready no longer blocks a child on a coordination parent's
non-existent branch.
#19: a task left claimed/in_progress with an assignee but no running container
was invisibly stuck (only PENDING tasks get fresh dispatch; the heartbeat reaper
can't see a freshly-seeded claim). New _dispatch_claimed_without_agent net:
after a short grace window it respawns the assignee, or releases the claim to
pending (lifecycle-safe via unclaim_for_reaper) when the assignee is unknown.
New config ROBOCO_CLAIMED_NO_AGENT_GRACE_SECONDS (default 120).
* fix(gateway): tolerant note verb + lock evidence do-tool invariant
#15: the note verb no longer hard-rejects thin decision/reflect payloads.
List-typed fields (options, consequences, next_steps) coerce a lone scalar
into a one-element list at both the NoteRequest schema (mode=before
validator) and the service layer; missing narrative fields default to a
visible placeholder instead of returning incomplete_input. The note is
always recorded, preserving audit value, and a well-intentioned note can no
longer trip the do-server 3-strikes circuit breaker. Widen the agent-facing
do_server.note hints to accept list-or-scalar and refresh the docstrings.
#8: add regression coverage locking the invariant that every role's do_tools
carries evidence (role_config + developer spawn manifest). The current source
already registers mcp__roboco-do__evidence for developers end-to-end; the
report stemmed from a stale deployed build, and the tests prevent silent
regression.
* fix(gateway): allow UX devs to receive design tasks; surface delegation rules to cell PM
The UX/UI cell's developers (ux-dev-1/ux-dev-2, Role.DEVELOPER on
Team.UX_UI) ARE its designers, but _validate_assignee_task_type rejected
task_type='design' for every DEVELOPER, blocking the UX cell's normal
design delegation. Allow 'design' for UX-team devs only; backend/frontend
devs stay rejected (design routing belongs to the UX cell). The
orchestrator already dispatches a developer for a design task
(_dev_dispatch_role_matches returns True), so this creates no orphan like
the documentation case.
Replace the static Cell-PM 'pass planning' remediate with a per-assignee
hint so a dev/design mis-type gets a developer-class next-step instead of
an off-topic planning hint.
Surface the three delegation guardrails in the cell-PM prompt so PMs stop
probing them by trial and error: valid task_type per assignee (incl.
design for UX devs), documentation auto-creation (non-delegatable), and
the sequential single-active code-spine. Fix the delegate-row task_type
list (documentation is NOT delegatable) and update the lifecycle spec
description; regenerate the lifecycle artifacts.
* fix(orchestrator): improve agent briefings for handoff consumption, product/project model, and workspace/secret hygiene
Main PM (roles/main_pm.md):
- Require reading the upstream Product Owner / Head of Marketing handoff
(their decision/reflect journal entries + task description) BEFORE doing
any own research or calling i_will_plan, so the Main PM builds on the
Board's analysis instead of duplicating it. Added a dedicated section,
hardened workflow step 1, and added an anti-pattern.
- Add a 'Products vs Projects' section: a Product fans out to one Project
per cell; those Projects may be the SAME repo (monorepo subtrees) or
DIFFERENT repos (multi-repo). The Main PM coordinates across them and
must not assume one repo or call a monorepo subtree 'a separate repo'.
Names the Prompter monorepo case (github.com/rennf93/roboco).
Developer (roles/developer.md):
- State the exact workspace path convention
/data/workspaces/<project-slug>/<team>/<agent-slug>/, that the cwd is
already set there, to stay inside the own cell workspace, and to not
probe/guess the path (ls /, find /).
- Sanctioned secret handling: env/printenv is bash-guard denied and
reveals nothing; needed secrets arrive via the task description, else
i_am_blocked so the PM supplies them. Added matching anti-patterns.
Tests: add tests/unit/agents/test_briefing_cluster_c4.py asserting the
composed system prompt (the text mounted into agent containers) carries
each of the above.
* fix(orchestrator): board review involves PO+HoM and notifies CEO
Cluster C5 (#2, #4): a board/coordination task was reviewed by the Product
Owner alone, and the CEO got no formal signal when the review finished —
only buried channel chatter — so the Approve & Start handoff was invisible.
#4 — Board review is now a two-reviewer gate. _handle_board_assigned_task
dispatches BOTH the Product Owner and the Head of Marketing (one-shot each),
regardless of which one holds assigned_to, and the unassigned board-routing
path delegates here instead of claiming + spawning the PO alone. Board tasks
stay pending/unassigned for the CEO's Approve & Start. The board prompt now
makes the PO+HoM pair-review model explicit (HoM owns the UX/positioning
dimension).
#2 — Once BOTH reviewers have finished (dispatched and no longer active),
the orchestrator emits exactly one formal CEO notification via
NotificationService.send_board_review_complete_notification (APPROVAL type,
ack-required, carrying related_task_id) so the handoff is an actionable
signal. One-shot per task; a notification failure clears the guard so a
later tick can retry.
To let the non-assignee board member record its review note on a task held
by the other board member, content-action ownership now exempts a board role
posting to a board/coordination task (project_id is None, product_id set).
The exemption is narrow: it does not widen ownership for any other role or
any project-backed task.
Unit tests cover both reviewers dispatched, one-shot dispatch, the CEO
notification fired exactly once when both are done (and not before), the
retry-on-failure path, the notification builder, and the board co-review
ownership exemption (allowed for board+coordination, blocked otherwise).
* fix(workspace): install dev deps post-clone + raise git commit timeout for large changesets
Cluster C6 (#10, #13, #12-investigate).
#10: per-agent workspace clones never had the project's dev dependencies
installed, so the make-quality gate (ruff/mypy/pytest for Python, the TS
toolchain for the panel) was missing and devs re-downloaded tooling per
task. WorkspaceService now runs the project's install after cloning
(`uv sync` for Python, `pnpm install`/`npm ci`/`npm install` for Node/TS,
detected by manifest/lockfile). Idempotent via a lockfile-digest marker
under .git/ so a re-entry with unchanged lockfiles is a no-op; also runs on
the healthy short-circuit so pre-existing clones get backfilled. Gated by
workspace_install_dev_deps (default on) with workspace_dep_install_timeout_seconds.
#13: the gateway commit verb timed out on the large panel changeset because
every git op used the hardcoded 30s _GIT_TIMEOUT and each call also re-walks
the tree to chown. _run_git now takes a per-call timeout override sourced
from settings (git_command_timeout_seconds default); the staging + commit
ops in commit() and create_commit() use the longer git_commit_timeout_seconds
(default 180s). httpx REST timeouts unchanged in value.
#12 (investigate only — no push, no history change): the clone base ref is
NOT hardcoded; it already comes from project.default_branch threaded through
git.get_workspace -> ensure_workspace -> _clone_repo (git clone --branch).
The stale-base problem is a deploy/process issue (GitHub master is behind the
deployed migration chain), resolvable only by pushing the chain to master.
The default_branch column is the existing configurable lever.
* fix(panel): gate Approve & Start to board coordination tasks; stop 404 storm on closed sessions
CEO gate #1 button only renders for a PENDING board coordination/fan-out
task (no project_id, has product_id) — the board-reviewed handoff that
approve_and_start accepts — instead of every PENDING board-team task.
approve_and_start requires PENDING (it re-targets to Main PM without a
status change), so the gate stays on PENDING rather than the unrelated
end-of-work awaiting_ceo_approval state.
Session/message reads now treat a 404 as terminal and never retry it: a
reaped session is gone for good, and retrying every dead session-id is
what produced the growing 404 storm on GET /api/messages. The transcript
loads once (staleTime Infinity, no focus/reconnect refetch) so closed
sessions stay viewable without re-polling.
* fix(orchestrator): role-correct respawn prompt, throttle agentless dispatch, broaden #14 guard
#19 wrong-role prompt on respawn: _get_prompt_for_agent fell through to the
developer prompt for every non-dev/doc/qa role, so a respawned PM or board
agent was told to write code and call verbs it does not own. Route by the
agent's actual role through the existing per-role prompt builders
(developer/qa/documenter/cell_pm/main_pm/product_owner/head_marketing/auditor).
Both callers benefit; _spawn_pending_dev only ever passes developer/documenter/
unknown, so its behavior is unchanged.
#19 spawn-burst: _dispatch_claimed_without_agent looped over every agentless
claimed/in_progress task and could spawn many containers in one tick. Break
after the first respawn so a restart can't trigger a burst, matching every
sibling dispatcher. The release-to-pending path spawns nothing and keeps
draining stale unknown claims.
#14 guard scope: _is_descendant_code_task only matched CODE, so a descendant
DOCUMENTATION or DESIGN task escalated to a board/advisory role was still
stranded on a role with no verb to own it. Rename to
_is_descendant_executable_task and broaden to CODE/DOCUMENTATION/DESIGN — the
cell-executed types a board role cannot own. PLANNING/RESEARCH/ADMINISTRATIVE
route to a PM, not a cell agent, and are left unchanged; root tasks are still
reviewed up the chain.
* fix(docker): add node+pnpm to orchestrator so it pre-installs frontend cell deps
* Added .github workflows
* refactor(services): extract helpers to keep install_dev_deps + developer task-type check under the xenon complexity gate
* chore(github): add launch kit — CI, GHCR release, labels, templates, funding, dependabot npm, community docs
* chore(github): bump_version — drop unused noqa, fix datetime UTC import
---------
Co-authored-by: Renn F <rennf93@users.noreply.github.com>
|
||
|
Renn F
|
e3def6b3a2 |
fix(gateway): cell PM completes its own cell task; drop main-PM handoff
submit_up bubbled the cell task to Main PM (_handoff_to_main_pm), but
main_pm_complete rejects any task with a parent_task_id ("only operates
on root tasks"), so the cell->root PR had no one to merge it and the
cell task wedged at awaiting_pm_review. _maybe_advance_parent_to_pm_review
already intends the CELL PM to complete it.
Cell PM now owns cell completion:
- submit_up no longer hands off to Main PM; the cell task stays assigned
to the cell PM, which is respawned to complete() it. Removed the
now-unused _handoff_to_main_pm.
- cell_pm_complete resolves the merge target from the parent task's real
branch_name (shared merge_chain.resolve_parent_branch, also used by the
PR side-effects) so the cell->root PR merges into feature/main_pm/...,
not the team-mis-derived feature/<cellteam>/... (same root cause as the
prior PR-base fix).
- submit_up description + next_hint updated; lifecycle artifacts regen.
Main PM still only completes the ROOT (root->master + escalate-to-CEO).
First run to reach cell-PM bubble-up exposed this.
|
||
|
Renn F
|
3d34fc2677 |
feat(progress): plan-driven progress — % derived from the plan checklist (#173)
Progress was only the synthetic milestone entry (auto-emitted at open_pr/i_am_done); agents never deliberately reported and the % was an ungated free-form guess. Now the plan's sub_tasks ARE the progress skeleton: - progress() gains optional `plan_step` (a sub_task id or its 1-based order). With it, that step is marked completed and the percentage is DERIVED as completed/total (equal weight) via new TaskService.record_plan_progress — the agent cannot set/game it. - A narrative entry WITHOUT plan_step is allowed for important mid-step documentation and carries the current derived % (the bar never regresses). No hard anti-spam gate (would loop minimax) — prompt guidance steers "meaningful moments, not every tool call". - `percentage` is now an optional fallback, used only for tasks with no sub_task checklist (back-compat). v2 ProgressRequest, the do.py route, and the do_server MCP tool updated accordingly. - An unmatched plan_step returns invalid_state listing the valid step refs (resolve by id / order / 1-based index). - developer + documenter prompts updated to the plan_step workflow. - Helpers extracted (_plan_subtasks/_derive_plan_pct/_valid_step_refs/ _mark_subtask_complete) to keep record_plan_progress within the cyclomatic gate. Commit 3 of 3 for the plan/progress quality work (#171/#172/#173). |
||
|
Renn F
|
4c397e1768 |
feat(gateway): developer i_will_work_on takes a substantive step checklist (#172)
The dev plan was a free string with only a presence gate, so the executing dev had no checklist for plan-driven progress (#173). - IWillWorkOnRequest gains `steps` (same SubTask shape as a PM's sub_tasks); flow_dev route threads it through. - i_will_work_on layers steps onto the narrative plan via the same panel-shaped path PMs use, so task.plan.sub_tasks is populated (panel render + #173 progress). - New _dev_steps_gate (mirrors _pm_sub_tasks_gate, runs after the spec gate): a developer FRESH claim must supply a non-empty steps list with every description >= _PM_SUBTASK_DESC_MIN_LEN. Re-entry/recovery short-circuit before the gate (extracted _dev_reentry + _fresh_dev_claim keep i_will_work_on within the return-count + cyclomatic gates). - developer role prompt: steps template + "thin steps rejected" + the progress(plan_step=...) handoff. - Updated every dev-fresh-claim test fixture across the suite to pass substantive steps; added dedicated _dev_steps_gate coverage. Commit 2 of 3 for the plan/progress quality work (#171/#172/#173). |
||
|
Renn F
|
ed828a719b |
feat(gateway): substantive-plan gate — approach >=150 + real sub_task descriptions (#171)
Plans were vague because the gate accepted the bare minimum: PM approach >=20 chars and title-only sub_tasks. minimax wrote exactly the minimum. - IWillPlanRequest.approach min_length 20 -> 150 (kept in sync with _PM_APPROACH_MIN_LEN; the gate enforces it at the choreographer layer too so direct/MCP callers can't bypass the HTTP boundary). - New _thin_subtask_hint: every PM sub_task must have a title and a description >= _PM_SUBTASK_DESC_MIN_LEN (60) saying what the step does — each sub_task is both a delegate target AND a progress-checklist item, so a title alone is not a plan. - cell_pm/main_pm role prompts: explicit "the gate REJECTS thin plans" framing + concrete sub_task example + the new minimums. - Updated all affected test fixtures across the suite to use substantive approaches/descriptions; added thin-sub_task rejection coverage. Commit 1 of 3 for the plan/progress quality work (#171/#172/#173). |
||
|
Renn F
|
38dba74837 |
fix(agents): stop instructing agents to ToolSearch built-in tools (#167)
The system-prompt directive layer and the briefing block both opened
with "FIRST ACTION REQUIRED: run ToolSearch to activate deferred
Edit/Write". That premise is false: per Claude Code 2.1.114, ToolSearch
gates only deferred MCP tools, never built-ins — and it is not even a
callable tool in the agent runtime. Built-ins are loaded at spawn via
the `--tools` flag and gated solely by the per-role permission rules
(the actual Edit/Write breakage was the global Write(*)/Edit(*) deny +
single-slash path, fixed in
|
||
|
Renn F
|
64d89fbd93 |
docs(prompts): teach all roles the roboco-git-readonly verbs
Smoke-8: QA fell back to Bash for git inspection (git log, git branch, git show) — bash-guard correctly blocked most of it. The roboco-git-readonly MCP server WAS registered for every agent (per orchestrator.py:1897) with roboco_git_status/log/diff/branches, but no role prompt mentioned them, so agents never tried them. Added the four verbs to the verb tables in: developer.md, documenter.md, qa.md, cell_pm.md, main_pm.md, board.md. Each entry notes "use these, NOT raw `Bash git ...`" so agents reach for the right tool first. No code change — these MCP tools have existed all along. This is a prompt visibility fix. |
||
|
Renn F
|
ef29d663fa |
docs(prompts): teach PMs the gateway's auto-naming conventions
Smoke-8: QA correctly failed a PR because the PM wrote acceptance criteria the gateway can never satisfy: - "branch named feature/backend/<full-uuid>" — gateway generates hierarchical 8-char IDs with `--` separator - "commit prefix [<root-id>]" — gateway prefixes with the leaf (dev's) task ID, not the root The dev did the right work (timestamp added to README, PR opened) but the literal criteria were unreachable. Both cell_pm.md and main_pm.md now have an "How to write acceptance_criteria" block explaining: - Gateway-controlled outputs: branch name, commit prefix - Examples of bad criteria (implementation/identifier-based) and good criteria (outcome/file-content/PR-state) - When you must reference a task ID, use the leaf (dev) ID — not the root. Next smoke run: PMs should write outcome criteria, dev work should clear QA on first review (assuming the work itself is correct). |
||
|
Renn F
|
87b18bc64f |
fix(gateway): spine-cap remediate forbids task_type workaround
Smoke-7: be-pm got the expected spine-cap rejection on a second delegate. The remediate said "drive the existing sibling to completion / cancel it, OR split this parent into two sibling parents". The model read that, decided neither applied, and "adapted" by re-delegating with task_type='documentation' as a "verification subtask". The gateway accepted it (different type = no cap collision) but the orphan subtask had no claimant — it blocked submit_up forever with "subtasks not all terminal". Two-layer fix: 1. _spine_type_dup_envelope remediate now explicitly forbids the workaround: "DO NOT work around this by delegating again with a different task_type (e.g. 'documentation' or 'research' as a 'verification' subtask). The lifecycle handles QA, documentation, and PM-review automatically after the code subtask finishes — you do not create auxiliary subtasks for those roles. Call i_am_idle() now and wait for the existing child to come back." 2. cell_pm.md workflow step 6 strengthened to name the anti-pattern explicitly: no verification subtask (QA is the verification step); never re-delegate with a different task_type as a workaround. 3 new tests pin the remediate text: forbids workaround, names the verification anti-pattern, retains invalid_state error kind. |
||
|
Renn F
|
197b1576c3 |
fix(prompts): hoist ToolSearch activation to top of system prompt
Smoke-7: be-dev-1 hit "Edit exists but is not enabled in this context." Claude Code v2.1.69+ defers built-in tools (Edit, Write, Read, etc.) behind a ToolSearch call. Weak models (minimax-m2.7) skip soft directives buried in the briefing. Also: 4 role prompts (developer, cell_pm, main_pm, board) claimed "no ToolSearch needed" — a lie that compounds the problem. The manifest registers MCP tools; built-in tools are still deferred. Fix: compose_prompt now prepends a tool-load directive layer as the FIRST block in the system prompt. It names the exact ToolSearch call the role needs: - developer/documenter: Read, Bash, Grep, Glob, Task, TodoWrite, Edit, Write - qa/pm/board: Read, Bash, Grep, Glob, Task, TodoWrite (no Edit/Write) The directive includes the failure mode it prevents so the model understands what skipping the call causes. Updated role-prompt lines that lied about ToolSearch. 7 new tests pin: directive is the first block; developer/documenter get Edit/Write; qa/pm don't; failure-mode message is present. |
||
|
Renn F
|
b4fe0f13fa |
feat(roles): expose pr_update to dev/doc/cell_pm/main_pm + prompt updates
Adds pr_update to _DEV_DO, _DOC_DO, _CELL_PM_DO, _MAIN_PM_DO so the spawn manifest builder registers it on those roles' do-servers. QA, auditor, and Board roles do not get it — QA reviews PRs but does not edit them; Board operates above the PR layer; auditor is silent. developer.md and documenter.md grow a verb-table row and a note next to the open_pr workflow step calling out that pr_update — not bash- shimmed `gh pr edit` — is the way to fix PR metadata. |
||
|
Renn F
|
4f7dd7a336 |
docs(prompts): E4 clarify TodoWrite vs progress() distinction
TodoWrite is Anthropic's private session-local scratchpad — agents use it to track their own immediate next steps. It does NOT surface to the panel's Progress tab and is NOT a substitute for progress(task_id, message, percentage). Smoke run 3 didn't show this conflation yet, but Wave D's new progress() directive risks it. - base.md gets the canonical "TodoWrite vs progress()" callout - developer.md + documenter.md (the two roles with progress()) get inline reminders in their verb tables: "NOT TodoWrite" Spec ref: docs/superpowers/specs/2026-05-12-post-smoke-3-fixes-design.md section E4. |
||
|
Renn F
|
ce5761701d |
docs(prompts): D1 fixup — role-correct circuit-breaker escalation paths
QA / Documenter / Board don't have i_am_blocked in their manifests. The D1 snippet's "escalate via i_am_blocked" line is now role-correct: - QA / Documenter: unclaim(task_id) + dm(cell-pm, ...) with rejection - Board: dm(ceo, ...) for PO/HoM; Auditor uses note(scope='reflect', ...) |
||
|
Renn F
|
df993befe7 |
docs(prompts): D4 compel channels() before invented say() slugs
All role prompts now mention channels() as the way to list valid
channel slugs. Smoke run 3 showed agents inventing slugs ('backend-dev',
'backend') and getting Channel not found. The channels() verb was
added in Wave 2 G6 but unused — making the directive explicit in
every prompt.
|
||
|
Renn F
|
dc11ca23f8 |
docs(prompts): D4 compel triage() first on respawn in cell_pm.md
Cell PM prompt now mandates triage() as the first call on every respawn, before re-decomposing. Smoke run 3 showed PMs re-decomposing blindly and hitting spine-cap; triage shows them existing children and prevents the over-decomposition pattern. |
||
|
Renn F
|
517aa4b16b |
docs(prompts): D4 compel progress() after each commit in developer.md
Developer prompt now mandates progress(task_id, message, percentage) after each commit. Wave 1's progress verb was added but no agent called it. The Progress tab stays empty without it. |
||
|
Renn F
|
02c241e3a5 |
docs(prompts): D4 compel open_session in PM prompts
PM prompts now include open_session(task_id, channel, topic) in the State→Verb table for the "just claimed" state. Without this, the Sessions tab stays empty — Wave 1's session verb was added but agents never called it because the prompt didn't directive it. |
||
|
Renn F
|
129504a51d |
docs(prompts): D3 document journal:during_work>=1 in developer.md
Smoke run 3 showed be-dev-1 writing reflect but no mid-work entry, hitting tracing_gap on i_am_done with missing: ['journal:during_work>=1']. The reflect note does not satisfy this gate — it's an end-of-work artifact. The prompt now shows the 5-step cadence explicitly: i_will_work_on → decision → work → reflect → i_am_done. |
||
|
Renn F
|
68d52be4a7 |
docs(prompts): D2 post-first-delegate reasoning for Main PM + Cell PM
Smoke run 3 showed Main PM seeing the spine-cap reject on its 2nd delegate attempt (its 1st succeeded) and concluding 'I cannot delegate' → escalated to product-owner. The new anti-pattern tells PMs that spine-cap or role-guard rejections AFTER a successful delegate mean over-decomposition, not delegation impossibility — verify with triage() and idle instead. |
||
|
Renn F
|
314d829172 |
docs(prompts): D1 circuit-recovery instruction in all 6 role prompts
Smoke run 3 showed be-dev-1 hitting circuit_open on i_am_done and escalating via i_am_blocked instead of writing the missing journal entry and retrying. The prompts now name circuit_open explicitly, tell agents to read the remediate, fix the one piece, retry once, and only escalate if the breaker fires again. |
||
|
Renn F
|
dc9c49e1e4 |
feat(gateway): G8 part b — typed blocker_type + what_needed on i_am_blocked
Pre-gateway parity (G8 part b of the 2026-05-11 design). The pre-gateway TaskBlockInput at 254cc93:roboco/mcp/schemas/__init__.py required blocker_type (external|internal|question|dependency) and what_needed so PMs could triage their inbox by class. Current i_am_blocked dropped both fields — every blocked task looked the same to the PM. Now i_am_blocked accepts both as optional kwargs: - Back-compat: callers that omit them still work (blocker_type defaults to None → rendered as flat reason in the struggle entry). - New: when supplied, the struggle journal entry body is structured markdown (## Blocker Type / ## What Needed sections) so the panel's journal view renders named blocks instead of one flat sentence. Validator on blocker_type enforces the enum at the Pydantic boundary with a clear "must be one of: ..." error if the agent invents a value (same pattern as the Wave 3 G7 validators). G8 part a — typed `pause(checkpoint_summary, remaining_work)` — defers. That gap needs a new IntentSpec in foundation/policy/lifecycle.py (currently pause is an ActionSpec only; agents auto-pause via i_am_idle) plus checkpoint wiring through TaskService.add_checkpoint. Material work, deferred until after the user has deployed and verified G7 + G8b lands cleanly. Wired: - roboco/api/schemas/v2/flow.py — IAmBlockedRequest gains optional blocker_type + what_needed; @field_validator enforces the enum - roboco/api/routes/v2/flow_dev.py — passes the new fields through - roboco/services/gateway/choreographer/_impl.py — i_am_blocked signature + structured struggle-entry rendering - roboco/mcp/flow_server.py — typed wrapper with the kwargs - agents/prompts/roles/developer.md — updated verb table - tests/unit/mcp_servers/test_flow_server.py — updated to expect the new optional kwargs as None when omitted Quality: ruff + mypy clean. 505 tests pass. |
||
|
Renn F
|
8408d761ca |
feat(gateway): Wave 1 pre-gateway parity — sessions, progress, notify inbox
Closes empty-panel complaints (Sessions, Progress) and the i_am_idle notification-inbox deadlock identified in the 2026-05-11 gap analysis. All backend service methods already exist; this is pure MCP-surface widening on top of the existing choreographer + ContentActions. New MCP tools (roboco-do): - progress(task_id, message, percentage) — Progress tab writer - open_session(task_id, channel, topic, ...) — Sessions tab writer (PM+) - link_session(session_id, task_id, ...) — Idempotent task↔session - notify_list(unread_only, pending_ack_only, limit) - notify_get(notification_id) - notify_ack(notification_id) Wired through: - roboco/api/schemas/v2/do.py — six new request schemas with Field constraints (Progress.percentage: ge=0, le=100; OpenSession.topic: max_length=200; etc.) - roboco/api/routes/v2/do.py — six new POST routes, thin dispatchers - roboco/services/gateway/content_actions.py — six new ContentActions methods forwarding to TaskService.add_progress, MessagingService.create_session_for_tasks /link_session_to_task, NotificationDeliveryService.list_for_agent / get_for_ recipient_and_mark_read / acknowledge - roboco/mcp/do_server.py — six new typed tool wrappers + registered in _TOOLS - roboco/services/gateway/role_config.py — receivers (list/get/ack) added to every role except auditor (who gets list/get, no ack). Session verbs to PM-or-up. Progress to dev + doc. - agents/prompts/roles/*.md — verb tables updated for developer / QA / documenter / cell_pm / main_pm. i_am_idle line points to notify_list as the deadlock resolution path. Authorization: - progress: assignee + active status (in_progress / verifying / awaiting_qa / awaiting_documentation) - open_session: cell_pm / main_pm / product_owner / head_marketing / ceo - link_session: caller must own the task - notify_ack: caller must be a recipient (ValueError from service maps to not_authorized envelope) Per-file ignore extended: - roboco/services/gateway/**/*.py = [PLC0415, PLR0913] — same rationale as roboco/mcp/**: typed verb signatures are the agent-facing contract; bundling into dataclasses hides field-level schema the LLM needs at the tool layer. Quality: ruff + mypy clean. 503 unit tests pass on touched surfaces. Spec ref: docs/superpowers/specs/2026-05-11-pre-gateway-parity-design.md |
||
|
Renn F
|
6369184b72 |
docs(prompts): mark i_will_work_on plan param as required, not optional
Smoke run showed be-dev-1 repeatedly calling `i_will_work_on(task_id)` without `plan`, hitting `tracing_gap missing=['plan']` and retrying with the same payload. Root cause: prompt's verb table showed the signature as `plan=None` (optional default) while the gateway requires plan on every claim — including first claim. The dev followed the signature line, missed the workflow-table line that pairs it with `plan='...'`. Tightening the signature to `plan` (no default), explicit "REQUIRED even on first claim" callout, and a note that resume calls use `plan='resume: <next step>'`. |
||
|
Renn F
|
92badfe6a0 |
docs(prompts): teach all roles the structured verb shapes (pre-gateway parity)
Counterpart to |
||
|
|
207aaecd72 |
Feature: lifecycle canonical spec (#14)
* chore: clean make quality baseline on feature/lifecycle-canonical-spec
Three classes of pre-existing issues blocking `make quality`:
1. Alembic migrations 002/009/011 used runtime introspection
(op.get_bind() + inspect / bind.execute) without guarding for
offline (--sql) mode. `alembic upgrade head --sql` is part of
`make quality`; in offline mode `op.get_bind()` returns a
MockConnection with no inspection system, so the migrations
crashed before emitting their SQL stubs. Each migration now
short-circuits or simplifies in `context.is_offline_mode()` —
live-DB behavior is unchanged.
2. ruff format drift on three files left over from prior in-flight
edits (choreographer/_impl.py, content_actions.py, and one test
file). `ruff format` applied.
3. vulture flagged two unused `tb` parameters in async __aexit__
stubs in test_task_service_lifecycle_misc.py. The parameter is
protocol-required but unused by the body — renamed to `_tb`
(vulture treats underscore-prefixed names as intentionally unused).
`make quality` is now green from this branch's HEAD; subsequent
lifecycle-spec work can use it as the per-task gate.
* feat(lifecycle): canonical spec package + Role/Status/TaskType enums
Foundation for the canonical lifecycle/permissions module. Enums
mirror docs/internal/old/workflows/STATUS_TRANSITIONS.md +
PERMISSIONS.md. Tests pin enum membership against both the
predecessor canon and roboco.models.base.TaskType.
* feat(lifecycle): Decision dataclass with allow/reject/tracing_gap constructors
Single rejection shape every consumer maps to its native format
(Envelope, HTTP code, prompt hint). __post_init__ enforces the
allowed/rejection_kind invariants so a malformed Decision can't reach
a consumer.
* fix(lifecycle): tighten Decision invariants per Task 2 review
Two reviewer findings on the Task 2 Decision dataclass, addressed
in one commit:
1. The docstring promised `allowed=True ⇒ rejection_kind is None
AND missing == [] AND remediate is None`, but __post_init__ only
checked the rejection_kind half. A caller could construct an
allow-shaped Decision with stale missing/remediate fields and
sneak it past validation. Tighten __post_init__ to enforce the
full invariant. Add a regression test.
2. tracing_gap defensively copies the missing list (`list(missing)`)
to isolate the stored list from later caller-side mutation, but
no test pinned this. Add a regression test that mutates the source
list after construction and asserts the stored list is unchanged.
Issue 2 from the same review (mutable list vs tuple for `missing`)
is a broader design call deferred until consumers exist; the
defensive copy is sufficient until then.
* feat(lifecycle): Precondition/ActionSpec/IntentSpec/StatusTransition dataclasses
The four dataclasses that hold the canonical tables. ActionSpec and
StatusTransition are direct ports of pre-gateway PERMISSIONS.md +
STATUS_TRANSITIONS.md rows. IntentSpec is the gateway-only addition:
each gateway intent verb declares which atomic actions it composes.
* feat(lifecycle): _STATUS_TRANSITIONS table + STATUS_GRAPH view
Direct port of STATUS_TRANSITIONS.md. Every transition records its
trigger action and (optionally) a role constraint. STATUS_GRAPH is
the precomputed source→{targets} view callers use for reachability
checks.
* fix(lifecycle): pin role_constraint values + clarify Task-5 handoff
Two reviewer findings on Task 4 _STATUS_TRANSITIONS, addressed in
one commit:
1. The original Task-4 tests verified (source, target) pairs but
not role_constraint contents. A typo in a single role name (e.g.
forgetting MAIN_PM from escalate_to_ceo) would have slipped past
them silently. Add test_status_transitions_role_constraints_match_canon
pinning every non-None constraint and the cancel-block invariant.
2. role_constraint=None on the `claim` rows from PENDING and
NEEDS_REVISION was load-bearing — it is the explicit handoff
point between the StatusTransition table (state machine layer)
and CLAIM_RULES (per-role claim authority, lands in Task 5).
The original inline comment said this in passing; expand it so
the design choice is unmissable for a stranger reading just
spec.py.
* feat(lifecycle): _ATOMIC_ACTIONS + CLAIM_RULES + ROLE_TEAM_RULES tables
Direct port of PERMISSIONS.md. Every task management tool gets an
ActionSpec with allowed_roles, source_statuses, target_status,
self_review_block, and needs_team_match flags. CLAIM_RULES maps each
Role to the statuses they can claim from. ROLE_TEAM_RULES is the
per-slug team restriction.
* fix(lifecycle): tighten ActionSpec contracts per Task 5 review
Three reviewer findings on Task 5's _ATOMIC_ACTIONS table, addressed
in one commit:
1. set_plan.source_statuses widened to {CLAIMED, IN_PROGRESS} but
every existing caller (i_will_work_on / i_will_plan compositions)
runs set_plan while CLAIMED, between claim and start. Narrow to
{CLAIMED} only. If a future "edit plan mid-flight" feature lands,
widen explicitly with test coverage at that time.
2. needs_team_match was set True only on claim/qa_pass/qa_fail/
docs_complete. Defense-in-depth says every role-scoped task
action should re-assert team match (don't rely on the inheritance
chain through assigned_to alone). Flip to True on: start,
set_plan, block, pause, submit_verification, submit_qa,
submit_pm_review, complete, create_subtask. Leave False on
board/CEO actions and PM cross-cell interventions (unblock,
resume, cancel) where the cross-cell semantics are intentional.
3. claim.source_statuses is intentionally a SUPERSET of any single
role's CLAIM_RULES allowance (the table holds the union; CLAIM_RULES
holds the per-role authority). Add an inline comment above the
claim ActionSpec so a future reader doesn't conclude the two
tables disagree — they don't, they encode overlapping facts at
different grains.
* feat(lifecycle): _INTENT_VERBS table — every gateway verb declared
Each gateway intent verb is now a named composition of atomic actions
plus optional side effects. i_will_work_on = (claim, set_plan, start);
i_am_done = (submit_verification, submit_qa); open_pr is pure side
effects (push_branch, create_pr); etc.
* fix(lifecycle): widen block.allowed_roles to include QA + Documenter
Task 6 review caught a role-set inconsistency: i_am_blocked.allowed_roles
admits dev/QA/doc, but the underlying block.allowed_roles only allowed
dev+PM. Result: a QA or documenter calling i_am_blocked would pass the
IntentSpec gate and then be rejected by the composed ActionSpec gate
when Task 7 wires can_invoke_intent.
Widen block to include QA + Documenter. The semantic case is sound: a
QA reviewing a task can discover an external blocker; a documenter
writing docs may need PM intervention. Predecessor PERMISSIONS.md
restricted block to dev+PM, but with the gateway exposing i_am_blocked
to all worker roles, the underlying atomic must agree.
The deeper unclaim/escalate_up "imperative verb" concern from the same
review (composes=() but mutates state) is deferred to Task 8 where the
validator design lands.
* feat(lifecycle): public lookup functions + Context + preconditions
can_claim, can_invoke_action, can_invoke_intent, valid_next_verbs,
composed_actions_for, intents_for_role, status_after — the entire
public surface every consumer will use. Context carries the
caller-supplied state preconditions need (plan, journal-decision
flag, etc.). Preconditions for plan/commits/no_pr/ownership are
declared once and wired into the relevant IntentSpecs.
* fix(lifecycle): wire PRECONDITION_OWNERSHIP through Context.actor_id
Task 7 review found _p_owns_task reads agent.id but every call site
passes None for the agent arg. Result: getattr(None, "id", object())
returns a fresh sentinel, task.assigned_to == <sentinel> is always
False, and open_pr / i_am_done would reject every owner the moment
Task 9 wires consumers.
Fix: thread identity through Context.actor_id (new UUID field) and
rewrite _p_owns_task to read from the context. Both call sites already
pass the Context — no signature changes elsewhere. Add green-path
test exercising the owner-can-open-pr case the existing tests
missed (the Task 7 plan only tested precondition-failure paths,
which masked the bug).
Plus surface hygiene: STATUS_GRAPH, CLAIM_RULES, ROLE_TEAM_RULES,
and the four PRECONDITION_* constants are now in
roboco.lifecycle.__init__.__all__ so consumers in Tasks 8/9 don't
depend on the implicit `from roboco.lifecycle.spec import ...`
backdoor.
* feat(lifecycle): import-time self-consistency validators
10 validators run at module import; first failure raises
LifecycleSpecError and prevents the package from loading. Covers
status enum coverage, reachability, terminal exits, intent
compositions, status chain consistency, claim-rule role/status
coverage, self-review symmetry, team-rule slug existence, and
StatusTransition action references.
* fix(lifecycle): close validator gaps; resolve BACKLOG-claim and submit_qa IN_PROGRESS-shortcut ambiguity
Three reviewer follow-ups on Task 8's _validate.py, plus two real
data corrections the new action-target-reachability validator
surfaced.
1. Design spec §9 calls for "every ActionSpec.target_status, when
set, is reachable from each source_status via STATUS_GRAPH" —
missing from Task 8's 10 validators. Add
_check_action_target_reachable_from_source.
2. _check_role_team_rules_slugs verified slug existence in
AGENT_UUIDS but NOT that the cell team in ROLE_TEAM_RULES
matches the seed. Add _check_role_team_rules_team_match,
scoped to non-None entries only — None means "exempt from
team-match enforcement" (cross-cell roles), not "no team in
org chart".
3. test_validators_pass_on_real_spec was ceremonial. Add
test_run_all_validators_raises_on_unknown_intent_action,
a deliberate-break regression that monkeypatches _INTENT_VERBS
to inject a fake action and asserts LifecycleSpecError raises.
The new action-target-reachability validator caught two real
data inconsistencies between the predecessor canon docs and the
spec tables:
A. claim.source_statuses listed BACKLOG and CLAIM_RULES[*PM]
listed BACKLOG, but STATUS_GRAPH[BACKLOG] = {PENDING, CANCELLED}
only. Resolution: PMs use the explicit \`activate\` action to
move BACKLOG → PENDING, then claim from PENDING. Drop BACKLOG
from claim.source_statuses and CLAIM_RULES.
B. submit_qa.source_statuses listed IN_PROGRESS, but
STATUS_GRAPH[IN_PROGRESS] does NOT include AWAITING_QA. The
intent verb i_am_done composes (submit_verification, submit_qa)
which forces IN_PROGRESS → VERIFYING → AWAITING_QA — no
shortcut. Drop the stale IN_PROGRESS entry from
submit_qa.source_statuses.
Both corrections tighten the canonical state machine to a strict
no-skip transition graph. Pre-gateway PERMISSIONS.md/STATUS_TRANSITIONS.md
disagreements are resolved here; spec.py is the canon now.
* feat(gateway): Envelope.from_decision maps lifecycle Decisions to envelopes
Single shape adapter so verb bodies stop hand-composing rejection
envelopes. Each rejection_kind maps to a specific envelope flavor;
'self_review' folds into 'not_authorized' with a parenthetical hint;
constructing from an allow Decision raises (programmer error).
* feat(gateway): VerbRunner for atomic composed-action dispatch
Wraps spec.composed_actions_for(intent) in session.begin_nested()
so mid-sequence failures roll the DB back. Side effects run AFTER
the savepoint commits. Each atomic action name dispatches to a
TaskService method via a single, exhaustive _dispatch_atomic
mapping. New verbs slot in by adding an IntentSpec entry + a
_dispatch_atomic case if a new atomic is needed.
* refactor(gateway): i_will_work_on uses spec.can_invoke_intent + VerbRunner
Replace the bespoke status-branch dispatcher in i_will_work_on with the
spec-driven flow: load task -> load agent -> build spec.Context ->
spec.can_invoke_intent (and spec.can_claim for per-role status authority)
-> Envelope.from_decision on rejection -> VerbRunner.run_intent on success.
The _i_will_work_on_pending, _i_will_work_on_claimed,
_i_will_work_on_needs_revision, and _start_failed_envelope helpers are
removed; the runner replaces them. Two narrow verb-body re-entry blocks
remain for behaviors the spec does not yet model:
1. in_progress + same agent -> idempotent heartbeat-only return
2. claimed + same agent -> _resume_from_claimed (set_plan + start)
to recover from a stuck mid-claim crash without re-running claim
against a state the spec excludes.
The behavioral claim guards (already_active / paused / sibling_sequence)
also stay imperative for now -- they're not in the spec yet and migrate
into spec.extra_preconditions in a later task. Per-role claim authority
is enforced via spec.can_claim because the atomic claim action's
source_statuses are the union across roles; CLAIM_RULES narrows.
Parity test in tests/lifecycle/test_consumer_parity.py runs the verb
against every (role x status x task_type='code') combo (112 rows) and
asserts the envelope error matches the spec's Decision (or can_claim's
Decision when the intent gate passes but per-role claim authority does
not). This is the contract that makes spec/verb drift impossible.
Existing tests updated where rejection-message text changed (the spec
now produces the messages, e.g. "role 'cell_pm' may not call
'i_will_work_on'" instead of "PM cannot execute code") or where the
spec's stricter view ("invalid_state" -> "not_authorized" for a dev
trying to claim awaiting_qa) is more accurate. Test fixtures were
updated to wire task.session.begin_nested as a proper async context
manager (required by VerbRunner) and to set agent_for().id so runner-
driven calls line up with assert_awaited_with(task_id, agent_id).
* refactor(lifecycle): push CLAIM_RULES enforcement into can_invoke_action
Task 11's i_will_work_on migration had to call spec.can_claim()
separately after spec.can_invoke_intent() because the claim action's
source_statuses is the union across all claim-eligible roles —
can_invoke_intent alone would let a developer pass for claiming
awaiting_qa (a QA-only state).
The retrofit pattern would repeat in every claim-composing verb
(i_will_plan, claim_review, claim_doc_task). Push the per-role
narrowing inside can_invoke_action when the action is "claim",
using the same not_authorized vs invalid_state disambiguation
can_claim already implemented (status-reserved-for-another-role
returns not_authorized; status-no-role-can-claim returns
invalid_state). Extracted the body to _check_claim_rules_narrow
to keep can_invoke_action under xenon's complexity threshold.
Update _i_will_work_on_gate to drop the redundant spec.can_claim
call. Update test_consumer_parity.py to assert only against
can_invoke_intent's Decision.
Tasks 12-22 will inherit the cleaner pattern: spec.can_invoke_intent
is the single gate; verb bodies don't need per-action retrofits.
* refactor(gateway): i_will_plan uses spec.can_invoke_intent + VerbRunner
Migrates i_will_plan to the spec-driven pattern Task 11 set up for
i_will_work_on. The verb body now: (1) loads task + agent, (2) builds
Context, (3) checks idempotent/recovery re-entry, (4) calls
spec.can_invoke_intent, (5) returns Envelope.from_decision on
rejection, (6) delegates composition to VerbRunner. The
_i_will_plan_* helpers are removed — the runner replaces them.
Parity test in tests/lifecycle/test_consumer_parity.py runs the verb
against every (role × status × task_type) combo and asserts the
envelope matches spec.Decision.
* refactor(gateway): delegate uses spec.can_invoke_intent for role/state gate
Migrates delegate to the spec-driven role/state gate. The chain
validation (main_pm->cell_pm, cell_pm->its team's devs), the
assignee-vs-task_type rule (Cell PMs receive planning-typed only),
the enum coercion, and the parent-lifecycle/cap guards STAY in the
verb body — they encode delegate-specific semantics the spec
doesn't model.
Parity test in tests/lifecycle/test_consumer_parity.py asserts the
spec's role+state rejection is correctly surfaced. Chain/assignee
rejections continue to be tested in test_choreographer_pm_extras.
* refactor(gateway): open_pr uses spec.can_invoke_intent + VerbRunner
Migrates open_pr to spec-driven gating. The spec's
extra_preconditions (PRECONDITION_OWNERSHIP, PRECONDITION_COMMITS,
PRECONDITION_NO_PR) handle all three precondition checks; the verb
body delegates side-effect dispatch (push_branch, create_pr) to
VerbRunner.
Idempotent re-entry retained: an open_pr call against a task that
already has a PR (and the caller owns it) returns OK without
re-opening, rather than the tracing_gap the spec would otherwise
produce. This preserves agent ergonomics — two calls in a row
shouldn't surface a misleading "no_prior_pr" hint.
Parity test in tests/lifecycle/test_consumer_parity.py runs the verb
against representative (status x commits x pr_number) combos and
asserts the envelope matches spec.Decision.
* refactor(gateway): i_am_done uses spec.can_invoke_intent + VerbRunner
Migrates i_am_done to spec-driven gating. The spec's
extra_preconditions (PRECONDITION_OWNERSHIP, PRECONDITION_COMMITS)
handle ownership and commit-count checks; VerbRunner dispatches
the (submit_verification, submit_qa) atomic chain.
The tracing-gate preconditions (progress entry, journal:reflect,
acceptance criteria) and the field-level submit-qa gates stay in
the verb body — they model gates the spec doesn't yet cover.
Defense-in-depth: those gates run after the spec accepts the
ownership/commits checks.
Parity test in tests/lifecycle/test_consumer_parity.py runs the
verb against (role × status × ownership × commits) and asserts
the envelope matches spec.Decision.
* refactor(gateway): i_am_blocked uses spec.can_invoke_intent + VerbRunner
Migrates i_am_blocked to spec-driven gating. The journal:struggle
write stays in the verb body (it's a side effect outside the
lifecycle action). VerbRunner dispatches the `block` atomic action
via task_service.escalate.
Parity test in tests/lifecycle/test_consumer_parity.py.
* refactor(gateway): unclaim and resume use spec.can_invoke_intent
Migrates both verbs to the spec-driven gate. unclaim's verb body
keeps its dispatch (task.unclaim_for_agent) because composes=();
resume goes through VerbRunner with composes=("resume",).
The reassignment-rejection branch (introduced in
|
||
|
Renn F
|
6806516015 |
refactor(gateway): rename submit_for_qa to open_pr; pin atomic preconditions
Pre-fix, submit_for_qa opened a PR (side effect) and returned OK with next='call i_am_done' — agents read the verb name, assumed they were done with QA handoff, never called i_am_done, and PRs ended up orphaned (PR #12 in the 2026-05-08 trace). Two changes: 1. Rename submit_for_qa -> open_pr so the verb name matches the semantic. The PR opens here; the actual QA handoff happens at i_am_done. Renamed across: - choreographer/_impl.py (method) - mcp/flow_server.py (tool registration + _TOOLS dict) - api/routes/v2/flow_dev.py (route + handler) - api/schemas/v2/flow.py (OpenPrRequest) - services/gateway/verb_gates.py (_STATE_VERBS) - services/gateway/role_config.py (developer flow manifest) - services/gateway/content_actions.py (commit-success next= hint) - agent_sdk/server.py (post-tool guidance map) - runtime/orchestrator.py (developer prompt) - agents/prompts/{base,roles/developer,_generated/*}.md - tests/unit/gateway/test_submit_for_qa.py -> test_open_pr.py - tests/unit/api/routes/v2/test_flow_dev.py - tests/unit/gateway/test_verb_gates.py - tests/unit/api/test_correlation_id.py - tests/unit/mcp_servers/test_flow_server.py - tests/integration/test_full_lifecycle_real_db.py 2. New regression test (test_open_pr_does_not_create_pr_if_no_commits) pins the atomic invariant: preconditions (assignee, commits, no-prior-PR) must be checked BEFORE git.create_pr/push_branch run. Any future re-ordering breaks the test. Tests: 3128 passing (3127 + 1 new), 100% coverage, ruff clean. Note: TaskService.submit_for_qa() (the v1-layer service method) is INTENTIONALLY not renamed — it's a different layer used by the v1 routes. The rename here is only the gateway verb surface. |
||
|
Renn F
|
85ef124c8f | Quality Gates |