qa
/qa
Uruchamia walidację etapu QA, generuje raport i działania po zakończeniu dla bieżącego zakresu, gdy etap QA jest gotowy do pętli weryfikacji.
- Autor
- grinrus ↗
- Repozytorium
- ai_driven_dev ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 08.09.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 6
- Skrypty
- 4
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 08.09.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
- 4 skrypty w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.
Skrypty: runtime/qa.py, runtime/qa_parts/__init__.py, runtime/qa_parts/core.py, runtime/qa_run.py
Co robi i kiedy
Uruchamia walidację etapu QA, generuje raport i działania po zakończeniu dla bieżącego zakresu, gdy etap QA jest gotowy do pętli weryfikacji.
Argumenty
/qa $1 [note...]
Oryginalny opis (EN)
Runs QA-stage validation, report generation, and postflight actions for the current scope. Use when QA stage is ready for loop verification. Do not use when the request belongs to `review` findings synthesis or `implement` execution loops.
Jak działa
- Steps
- Command contracts
- Notes
- Additional resources
Podgląd SKILL.md
---
name: qa
description: Runs QA-stage validation, report generation, and postflight actions for the current scope. Use when QA stage is ready for loop verification. Do not use when the request belongs to `review` findings synthesis or `implement` execution loops.
argument-hint: $1 [note...]
lang: en
prompt_version: 1.0.41
source_version: 1.0.41
allowed-tools:
- Read
- Edit
- Glob
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/qa/runtime/qa_run.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/qa/runtime/qa.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-docio/runtime/actions_apply.py *)"
- "Bash(rg *)"
- "Bash(npm *)"
- "Bash(pnpm *)"
- "Bash(yarn *)"
- "Bash(pytest *)"
- "Bash(python *)"
- "Bash(go *)"
- "Bash(mvn *)"
- "Bash(make *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-rlm/runtime/rlm_slice.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/set_active_stage.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/tasks_derive.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/progress_cli.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/status_summary.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/tasklist_check.py *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/tasklist_check.py --fix *)"
- "Bash(python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/set_active_feature.py *)"
model: inherit
disable-model-invocation: true
user-invocable: true
---
Follow `feature-dev-aidd:aidd-core` and `feature-dev-aidd:aidd-loop`.
## Steps
1. Inputs: resolve active `<ticket>/<scope_key>` and validate QA prerequisites from loop/readmap artifacts.
2. Stage-chain policy: execute via canonical stage-chain orchestration through stage runtime entrypoint; internal preflight/postflight are orchestration details and not operator commands.
3. Manual write/create of `stage.qa.result.json` is forbidden; stage-result files are produced only by stage-chain postflight. `[AIDD_LOOP_POLICY:MANUAL_STAGE_RESULT_FORBIDDEN]`
4. Runtime-path safety: use only runtime commands from this skill contracts. If stdout/stderr contains `can't open file .../skills/.../runtime/...`, stop with immediate BLOCKED `runtime_path_missing_or_drift`; one runtime-path error is terminal for current run, do not invent alternate filenames, and do not retry guessed commands.
5. Retry safety: do not rerun the same failing shell command more than once without new evidence/artifacts. For cwd/build mismatches, stop with blocker and handoff instead of looped retries.
6. Fail-fast convergence hardening: if preflight/stage-chain context is missing (`reason_code=preflight_missing`) or workflow paths resolve to non-workspace targets, stop immediately with terminal BLOCKED for the current run and explicit next action `/feature-dev-aidd:implement <ticket>`.
7. Actions contract hardening: if `actions-apply` reports schema/payload mismatch (`reason_code=contract_mismatch_actions_shape`), do not attempt guessed retries or manual payload edits; return terminal BLOCKED with the canonical handoff.
8. Read order after stage-chain preflight artifacts: `readmap.md` -> loop pack -> review pack (if exists) -> rolling context pack; do not perform broad repo scan before these artifacts.
9. Run subagent `feature-dev-aidd:qa`.
10. Orchestration: run QA via `python3 ${CLAUDE_PLUGIN_ROOT}/skills/qa/runtime/qa.py`, derive tasks if needed, then Fill actions.json for `aidd/reports/actions/<ticket>/<scope_key>/qa.actions.json` strictly as `aidd.actions.v1` (`schema_version`, `allowed_action_types`, canonical `type` + `params`) with action types only from `{tasklist_ops.set_iteration_done, tasklist_ops.append_progress_log, tasklist_ops.next3_recompute, context_pack_ops.context_pack_update}`, and validate via `python3 ${CLAUDE_PLUGIN_ROOT}/skills/qa/runtime/qa_run.py`. Do not use raw ad-hoc test commands from arbitrary cwd as a recovery path; run tests only via the QA runtime/tasklist contract.
11. Canonical stage-chain: internal preflight -> stage runtime -> actions_apply.py/postflight -> `python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/stage_result.py`. `[AIDD_LOOP_POLICY:CANONICAL_STAGE_RESULT_PATH]`
12. Non-canonical stage-result path under `skills/aidd-loop/runtime/` is forbidden (treat as prompt-flow drift). `[AIDD_LOOP_POLICY:NON_CANONICAL_STAGE_RESULT_FORBIDDEN]`
13. Output: return QA status contract with report paths and explicit canonical next action (`/feature-dev-aidd:status <ticket>` or `/feature-dev-aidd:tasks-new <ticket>` when follow-up tasks are required). Ensure one terminal payload per run (no repeated guessed recovery loops).
## Command contracts
### `python3 ${CLAUDE_PLUGIN_ROOT}/skills/qa/runtime/qa_run.py`
- When to run: as canonical QA stage runtime before postflight.
- Inputs: ticket, scope/work-item context, QA findings, and actions payload.
- Outputs: validated QA report artifacts and stage status payload.
- Failure mode: non-zero exit when report/actions schema or required stage inputs are invalid.
- Next action: fix QA findings/actions contract issues and rerun runtime validation.
### `python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-docio/runtime/actions_apply.py`
- When to run: mandatory final step in stage-chain postflight after QA actions are validated.
- Inputs: `--actions <path>` and optional `--apply-log <path>`.
- Outputs: applied actions plus progress/stage-result/status-summary artifacts.
- Failure mode: apply failure, progress update failure, or status summary failure.
- Next action: inspect logs, fix blocking contract issues, rerun the stage-chain.
### `python3 ${CLAUDE_PLUGIN_ROOT}/skills/aidd-flow-state/runtime/stage_result.py`
- When to run: stage-chain postflight stage-result emission only (not operator/manual recovery command).
- Inputs: canonical postflight payload (`ticket`, `stage`, `result`, `scope-key`, `work-item-key`, evidence links).
- Outputs: canonical `aidd.stage_result.v1` at `aidd/reports/loops/<ticket>/<scope_key>/stage.qa.result.json`.
- Failure mode: non-zero exit on missing required args or invalid stage-result contract fields.
- Next action: fix postflight payload generation and rerun the stage-chain; do not switch to non-canonical loop runtime paths.
## Notes
- QA stage runs full tests per policy.
## Additional resources
- Contract schema: [CONTRACT.yaml](CONTRACT.yaml) (when: QA action/report requirements are unclear; why: verify mandatory fields before runtime validation and postflight).
Pliki w paczce (6)
- CONTRACT.yaml 3 KB
- SKILL.md 6 KB
- runtime/qa.py 1 KB
- runtime/qa_run.py 1 KB
- runtime/qa_parts/__init__.py 1 KB
- runtime/qa_parts/core.py 33 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · grinrus/ai_driven_dev | 90 | MIT |
| anilcancakir/claude-code | 100 | MIT |
| builderio/agent-native | 100 | MIT |
| gethamster/plugin | 100 | MIT |
| paradigmxyz/centaur | 100 | MIT |
| lamngockhuong/aiteamkit | 100 | MIT |
| Everyone-Needs-A-Copilot/claude-copilot | 100 | MIT |
| adobe/skills | 100 | Apache-2.0 |
| svyatov/supermatt | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 2db66344505b (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 90implement
Wykonuje pętlę etapu implementacji dla kolejnego zakresowego zadania w orkiestracji etapów. - 100aidd-loop
Wymusza dyscyplinę trybu pętli dla etapów implement, review i qa: paczki, zakres, brak pytań i zasady ponowień. - 84plan
Prowadzi etap planowania w Fabriku: projektuje podejście do wdrożenia i tworzy checklistę zadań dla etapu implementacji. - 90verification
Prawdopodobnie prowadzi etap weryfikacji w potoku, gdy bieżący stan to verification, według instrukcji autora. - 90handoff
Tworzy krótki prompt startujący nową sesję na kolejny etap zakresu pracy, na koniec każdego etapu. - 90development-recipe
Prowadzi etap rozwoju zmiany: trzyma się zakresu, mapuje kryteria na testy, powtarza pętlę wewnętrzną i wstrzymuje się przy decyzji.