메뉴
Evidence-gated close-out for an item. Requires test output paste, code review, and quality-bar pass before squad done runs.
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
Every "tests pass" / "build green" / "feature works" claim needs the actual command output pasted into the conversation. Bare assertions are worth zero.
The operating loop every session follows — register, tick, pick, work, checkpoint, test, review, commit, close. Use whenever you start or resume work on a squad-managed repo.
Post often, post small, post honestly. The team needs visibility into your non-obvious state — not a change log. Use the typed verbs (thinking / milestone / stuck / fyi / ask) so peers can route attention.
Tick the squad team — surface mentions and file conflicts. Address mentions BEFORE continuing other work.
Before signing off — whether you finished an item or stopped mid-flight — post a 3-bullet summary. Shipped, in flight / queued, surprised by. Three bullets, no ceremony.
Run a structured intake interview with the user — open a session, ask one focused question per turn, draft the bundle, confirm with the user, then commit. Supports green-field (mode=new) and refine-mode (mode=refine) flows.
| name | squad-done |
| description | Evidence-gated close-out for an item. Requires test output paste, code review, and quality-bar pass before squad done runs. |
| argument-hint | <ITEM-ID> |
| allowed-tools | ["Bash","Read","Edit"] |
| paths | [".squad/items/**"] |
| disable-model-invocation | true |
You are closing item $ARGS. This command is evidence-gated — do NOT run squad done until every gate is green.
Pre-flight checklist (invoke each skill explicitly):
squad-evidence-requirement — Paste the actual output of every verification gate (tests, type-check, build, manual verification) into the conversation. Bare assertions do not count.squad-quality-bar — Walk the checklist over the diff: no commented-out code, no TODOs, no PM traces, no defensive checks for impossible cases, no half-finished work, AC literally checked off.squad-code-review-mandatory — Spawn superpowers:code-reviewer on the diff. Verify each finding (do not perform-agree). Address blocking findings before proceeding.Once all three gates are clean, run:
squad done $ARGS --summary "<one-line outcome>"
Learning capture (bug/feat/task only) — follow-through, not a precondition. After squad done runs, the CLI prints a one-line learning nudge tailored to the item type (gotcha-shaped for bug, generic for feature/task; silent for chore/tech-debt/bet). The lowest-friction capture is squad learning quick "<one-liner>" — auto-derives slug, defaults kind to gotcha, infers area from the most recently closed item, edit the stub later. Use squad learning propose <kind> <slug> when you want full control. Either way, file it while the surprise is still fresh — the Stop-hook prompt fires too late in multi-item sessions, by which time the detail has rotted. If nothing surprised you, ignore the nudge.
Then:
status: done, updated: to today, add ## Resolution section.squad-premise-validation..squad/done/.If you cannot meet a gate this session, set status: review instead of done, document what is missing, and stop. Do not ship past blocking gates.