Menu
Evidence-gated close-out for an item. Requires test output paste, code review, and quality-bar pass before squad done runs.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
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.