メニュー
Use when reporting that work is complete, fixed, passing, verified, ready for review, or safe to ship
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Multi-step implementation method to use when creating, studying or executing a plan. org-plan helper script.
Use when an implementation goal is complete enough to define coherent automated tests, or when implementation and tests must be brought to green before completion
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
Use when creating or editing reusable agent skills, improving skill discovery metadata, organizing bundled resources, or validating a skill package
SOC 職業分類に基づく
| name | verification-before-completion |
| description | Use when reporting that work is complete, fixed, passing, verified, ready for review, or safe to ship |
Match completion claims to current evidence. Report what was actually checked, with which command or inspection, and what result it produced.
When $org-plan is active, it controls verification boundaries, test scope, commits, L1/L2 progression, and continuation. This skill consumes the evidence produced by that workflow; it does not trigger an additional command, test run, pause, review, commit gate, or task-transition gate.
At an L2 boundary, use the touched-test and implementation evidence required by Org Plan. At an L1 boundary, use the full-suite evidence required by Org Plan. Do not broaden either boundary.
When reporting a result:
Do not rerun checks solely because a message is being written. Run checks only when the governing workflow, task, or user request requires them.
| Claim | Relevant evidence | Insufficient evidence |
|---|---|---|
| Touched behavior passes | Current focused-test result | An older run from before the change |
| Full suite passes | Current full-suite result at the required boundary | Focused tests only |
| Build succeeds | Current build result | Lint output |
| Bug is fixed | Reproduction or relevant test result | Code changed without checking behavior |
| Regression coverage is relevant | Assertions exercise the intended behavior | A passing test of incidental details |
| Requirements are met | Requirement-by-requirement inspection | Tests alone |
| Delegated work is complete | Inspect artifacts and relevant results | Agent report alone |
Focused tests: `pytest tests/test_widget.py` — 12 passed.
Full suite was not run at this L2 boundary.
The implementation is present, but no current test result is available; passing status is unverified.
Avoid broader statements such as “all tests pass” when only focused tests ran.
Use the evidence already required for the current workflow boundary. Make no claim broader than that evidence.