* fix(board): LEARN decisions name the item, not its per-cycle index A cycle's reject reasons are rendered into the NEXT cycle's exploration prompt, but the ref recorded alongside each reason was the item's stored id (item-0/item-1) — a per-cycle index that means something different every cycle and appears nowhere the explorer can resolve. The reason survived the loop; what it was about did not. Record the item's title instead, via a shared learn_ref() helper (falls back to the id when title-less, and reads target_task_title for Scales, whose items name the live task they mutate). * chore(lint): satisfy ruff 0.16 — keyword-only signatures and markdown formatting The dev toolchain resolved ruff 0.16.0, which stabilises PLR0917 (too many positional arguments) and formats python code blocks inside markdown. Both fired repo-wide and neither had anything to do with the code they flagged. - 36 signatures gain a `*` so their tail arguments are keyword-only, and the 104 call sites that passed them positionally are converted. mypy was the safety net for the static ones; the full suite caught nine more that only bind at runtime (the MCP tool functions, whose real callers already pass named JSON arguments). - 28 markdown files reformatted by 0.16's code-block formatter. - One RUF036 (`None` mid-union) autofixed in the GitLab provider. * fix(gateway): log the reason when a verb rejects A rejected envelope rides an HTTP 200, its body is never logged, and there is no trace table — so in the access log a verb an agent could not satisfy looks identical to one that worked. On 2026-07-25 four Board Programs (Periscope, Sentinel, Scales, Barfly) each POSTed their propose verb three or four times, persisted nothing, and left their exploration tasks PENDING; the reason was unrecoverable afterwards, from the logs or from the agents' own transcripts. Log error/message/remediate/missing plus the calling agent at envelope_to_response — the one chokepoint every v1 flow and do route returns through. Success envelopes stay silent. --------- Co-authored-by: Renn F <rennf93@users.noreply.github.com>
5.9 KiB
Pull Request Creation
When PRs Are Created
PRs are opened before QA review, not during awaiting_documentation. The choreographer creates the PR as a side-effect of the developer's open_pr(task_id) transition (verifying → awaiting_qa).
This is by design: QA reviews the real PR diff on the project's forge, and the downstream PM/CEO approval chain operates on a PR that already exists.
You do not call any tool to create a PR. There is no roboco_git_create_pr MCP tool.
How the dev triggers it
# 1. Make commits as you work (auto-pushes, no separate push step)
commit(
message="feat(api): add Redis rate limiter",
files=["roboco/api/routes/rate.py", "tests/integration/test_rate.py"],
)
# 2. Once acceptance criteria are implemented + tested, hand off to QA.
# The choreographer opens the PR here, sets pr_number/pr_url on the
# task, and transitions verifying → awaiting_qa.
open_pr(task_id="<task>")
The transition enforces (enforcement/task_lifecycle.py):
self_verified=True— set when you calli_am_done()orverify(task_id)firstcommitsnon-empty — at least one commit on the taskprogress_updatesnon-empty — at least one note on what changedpr_numberis set automatically by the choreographer; you don't pass it
If any precondition is missing, the verb returns an envelope explaining what's missing and how to remediate.
The push and the PR head always target the task's own branch by name, independent of whatever the shared clone happens to be checked out on. So a No commits between or wrong-branch worry at open_pr is the verb's job to resolve — never switch branches by hand to "fix" it.
PR Title and Body
Generated from templates in roboco/templates/git/pr_internal.py and roboco/templates/git/pr_root.py. You don't write the body by hand — it's filled with task title, acceptance criteria, the dev's notes, and the standard traceability links.
Title format: [TASK-{root-id:8}:{task-id:8}] {task-title}.
Parallel Documenter Phase
After QA passes, the task transitions to awaiting_documentation and runs documenter + dev in parallel:
| Agent | Action | Flag set |
|---|---|---|
| Documenter | Writes docs files, then i_documented(task_id, notes, files) |
docs_complete=True |
| Developer | (already done by the time we get here) | pr_created=True |
Task transitions to awaiting_pm_review when both are true.
PM Merges via complete
An assembled parent reaches awaiting_pm_review only after the in-path gate: the Cell PM's submit_up opens the cell→root PR and enters awaiting_pr_review, where the cell PR reviewer pr_passes it. The Cell PM then calls complete(task_id, notes). The choreographer:
- Verifies all subtasks are in a terminal state
- Verifies the PR is reviewable
- Merges the leaf PR into the parent branch (squash by default)
- Transitions the task to
completed
For the root parent, Main PM's submit_root opens the root→master PR and enters the same gate; after the main reviewer pr_passes it, the Main PM's complete escalates to the CEO (it does not merge). Only the CEO merges the root→master PR.
There is no roboco_git_merge_pr MCP tool.
Prerequisites
- Git token: the project must have an encrypted token set on
projects.git_token_encrypted. Without it, the workspace clone — and therefore everything downstream — fails withWorkspaceError. The field is historically named for GitHub PATs but works for any forge a project is registered against (GitHub, Gitea, GitLab —projects.git_provider); see "Forge-agnostic git" below. - Token scope:
repoon GitHub/Gitea; an equivalentapi/write_repositoryscope on GitLab (for branch push, PR/MR create, PR/MR merge). - Merge target: every dev/cell/root PR — never just the root→master one — targets the project's env-ladder head rung (
roboco.models.env_branches.head_branch, typicallymaster) — a project with no declared environment ladder resolves this straight fromprojects.default_branchvia the read-time shim, so this is unchanged for every project that hasn't opted into a multi-rung ladder. A middle ladder rung (e.g.qa/stag) is never a PR target for dev/cell/root work — the only thing that ever lands a PR on a middle or prod rung is theEnvSyncEnginecascade (seeCLAUDE.md"Env-branches ladder"), which is a platform-authored sync, not something you open.sync_branchandsubmit_upresolve their target from the task's own parent hierarchy (resolve_parent_branch), not the ladder directly — the ladder only surfaces at the two edge cases where a task has no branched ancestor: a project-root task's branch cut, andsubmit_root's PR target.
Forge-agnostic git
The PR/CI/review surface is provider-routed (roboco/services/forge/) — GitHub, Gitea, and GitLab are all supported per-project (projects.git_provider), and GitService never branches on which one a project uses. From your side, pr_number and pr_url are always real values regardless of forge — pr_url is the actual forge URL (a Gitea/GitLab instance host, never assumed to be github.com). One real asymmetry: GitLab has no "request changes" review primitive, so a pr_fail/change-request verdict on a GitLab-backed project posts as a plain MR note rather than a blocking review — the task still moves to needs_revision normally either way, only the PR-visible signal differs. Don't hardcode github.com in any URL you construct or reason about.
Troubleshooting
NO_COMMITSonopen_pr→ callcommit(...)first; nothing to open a PR over.NO_PRonpass/fail→ the choreographer didn't open a PR; check the workspace state withroboco_git_statusand re-callopen_pronce the workspace is clean.FORCE_PUSH_FORBIDDEN→ only the CEO may force-push. If your branch diverged,unclaimand re-claimthe task; the choreographer rebuilds the branch.