메뉴
When the runner is about to return -- hold the invocation open until the full PR lifecycle is done (push, mark ready, CI watch, merge). Returning early orphans the work.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when deciding whether to dispatch a runner or a worker for a subtask. Describes the decision boundary, the failure mode when runner is used where worker is intended, and the safe composition pattern.
When a repo has no GitHub remote (or is not yet a git repo) -- use this to route survey, sweep, and dispatch off the GitHub-centric path. Detect with `git remote` empty; treat blank GitHub columns as expected, not an error; route work to commits + CLAUDE.md instead of issues + PRs; degrade a maintenance sweep to the local-only axes.
When the directive is "status check," "what's new across my projects," or "sweep maintenance" -- run a per-project, read-mostly, non-PR sweep. For each project under the owner-prefixed layout, gather stars/forks, open issues/PRs split into mine vs community vs bots, package downloads, release-due (unreleased conventional commits), a test run, a stale-CLAUDE.md flag, and branch-cleanup candidates. Emit a fresh per-project report every sweep; NEVER cache status back into a workspace/manager CLAUDE.md.
Before making any non-trivial code, docs, or config change -- create a feature branch first, then open a PR. Never commit directly to main. Apply by default before any edit; this is the baseline branch discipline that all other git skills build on.
When the issue or PR queue has open unlabeled items -- run a read-only triage pass that labels each one by component, category, priority, and size, flags duplicates, closes noise, and reports the p1 queue before runners are dispatched.
When writing any issue, PR, or commit message, use this for the authoring standard. Covers structure (headings, lists, tables, fenced blocks, the bug block, the before/after) and voice (factual, no emdashes, no editorializing labels, conventional-commit prefixes, no commit trailers).
| name | runner-synchronous-lifecycle |
| description | When the runner is about to return -- hold the invocation open until the full PR lifecycle is done (push, mark ready, CI watch, merge). Returning early orphans the work. |
Your (the runner's) invocation must hold open until the full lifecycle is done. Returning to the dispatcher signals "task complete: PR is merged (or an exception was hit -- see exception cases in runner.md)." Returning earlier orphans the work.
Original failure mode observed on the work machine (roba #104):
run_in_background=trueThe dispatch is fired synchronously (no run_in_background).
Your session blocks until the dispatch exits.
# Mechanism per dispatch-options:
# Task tool: Task(subagent_type: "runner", prompt: <prompt>)
# Bash + roba: roba --fresh --full-auto -C <repo-path> -f /tmp/task-<N>.md
# Bash + claude -p: claude -p --agent runner "$(cat /tmp/task-<N>.md)"
For Bash-based dispatch, set a generous timeout (harness max is 600000 ms / 10 min; pick what fits the task size).
CI watch CAN use run_in_background=true because the watch is
part of your runner's lifecycle and YOU wait for the notification
yourself before returning. The
dispatch-wait-react skill is
the operational guide for that wait.
Push, mark ready, merge are all within your session. Don't hand them off to "the dispatcher will pick this up." The dispatcher's expectation is that when your invocation returns, the lifecycle is done.
When you DO return to the dispatcher:
needs-review or no-auto-merge label on the PR,
review:manual constraint, critical/delicate label), and "PR #N ready;
awaiting manual merge."The dispatcher's contract: "the runner returned" → "the lifecycle is complete." If you return earlier than that, you've broken the contract and the dispatcher will trust the wrong state.
Two PR labels signal the runner to skip auto-merge and return "awaiting manual merge" instead:
needs-review -- set by the dispatcher or issue author to require human
sign-off before merging.no-auto-merge -- set by the runner (copied from the issue) or manually,
to signal that automated merging should be skipped regardless of CI status.Check for these before calling gh pr merge:
LABELS=$(gh pr view $PR --json labels --jq '[.labels[].name] | join(",")' 2>/dev/null || echo "")
if echo "$LABELS" | grep -qE "needs-review|no-auto-merge"; then
echo "PR #$PR ready; awaiting manual merge (label: needs-review or no-auto-merge)"
# Do NOT merge
fi
dispatch-wait-react -- the
background + notification pattern for the CI-watch half of your
lifecycle.draft-pr-first -- the full
lifecycle your invocation must hold open through.runner-issue-authority --
the first step of that lifecycle (fetch the issue body).