/sprint-checkin

/sprint:checkin - Mark step done

Отмечает выполнение шага в active sprint.

Аргументы

• "<step description>" (обязательно): что было сделано. Specific, не vague.
• --evidence=<file> (опциональный): путь к файлу как доказательство (созданный документ, лог, etc.)

Использование

После каждого significant логического шага:

/sprint:checkin "Read 12 existing ADRs in docs/architecture/adr/"
/sprint:checkin "Wrote test for user_login function" --evidence=tests/test_auth.py
/sprint:checkin "Sprint goal achieved: architecture findings collected"

Что происходит

1. Validation

checkin-validator skill cross-checks:

• Does claim align with recent tool calls?
• Если evidence given - file exists и was modified recently?
• Если test claim - was test runner invoked?
• Heuristic checks for fake tests pattern

Warnings logged в .sprints/active/honesty-warnings.log. Не block workflow - только flag suspicious patterns для SE review.

2. State update

Update .sprints/active/_state.json:

• Last checkin timestamp
• Active task progress
• Step marked as done

Update .sprints/active/sprint-{N}.md:

• Append to checkin history
• Move step from todo to done

3. Tool history reference

Continue tracking tool history for future checks. No special action.

Special cases

Sprint goal checkin

Final sprint task должен иметь explicit goal achievement checkin:

/sprint:checkin "Sprint goal achieved: architecture findings collected"

Or partial:

/sprint:checkin "Sprint goal partial: architecture done, deployment phase deferred"

Or not achieved:

/sprint:checkin "Sprint goal NOT achieved: blocked by missing dependency"

DoD validator looks для one of these patterns.

HANDOFF.md update checkin

Mandatory before sprint close:

/sprint:checkin "HANDOFF.md updated with sprint outcomes"

handoff-keeper subagent validates что update реально произошло и meaningful.

Test-first checkin

Когда implement test before code:

/sprint:checkin "Wrote failing test for user authentication" --evidence=tests/test_auth.py

Это unblocks pre-tool-use-test-first hook для следующих production code changes.

Validation behavior

If alignment OK

Silent success:

✓ Checkin recorded

If warnings detected

Flagged but не blocking:

✓ Checkin recorded with warnings:
  ⚠ Test contains 'assert True' pattern in tests/test_auth.py:42
  ⚠ No test runner invocation in last 5 minutes

(SE reviewer will examine these warnings)

If no active sprint

Error:

✗ No active sprint. Use /sprint:start first.

Frequency

Recommended:

• After each logical step (5-15 minute work units)
• Before tool-intensive operations (so hooks don't BLOCK)
• After significant milestones

Don't do:

• Burst checkins ("done", "done", "done") - flag as suspicious
• Single checkin за весь sprint
• Vague checkins ("worked on stuff")

Output

Updated _state.json + sprint-{N}.md. Log in .sprints/active/honesty-warnings.log (если warnings).