Programowanie · Przegląd i audyt
audit
/audit
Przegląda zmianę kodu, plan albo wskazany moduł lub repozytorium i zgłasza ustalenia z oceną.
Dla kogo
- Autor
- nazmirket ↗
- Repozytorium
- skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 12.11.2026
Ocena i bezpieczeństwo
90/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść25 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Przestarzałość
Brak sygnałów. Repozytorium zmienione 12.11.2026, bez odwołań do wycofanych modeli i API.
Licencja
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Przegląda zmianę kodu, plan albo wskazany moduł lub repozytorium i zgłasza ustalenia z oceną.
Oryginalny opis (EN)
Use when asked to audit or review a code change, draft plan, or explicitly scoped module, system, or repository.
Jak działa
- Purpose
- Set scope and independence
- Code change
- Existing module, system or repository
- Draft plan
- Completion
Sekcje z nagłówków SKILL.md, w oryginale · 32 linii, 554 słów
Podgląd SKILL.md
--- name: audit description: "Use when asked to audit or review a code change, draft plan, or explicitly scoped module, system, or repository." --- # Audit ## Purpose Return prioritized, evidence-backed findings about the requested target. Audit is read-only unless fixes are authorized. [Test](../test/SKILL.md) owns behavioral checks; [Critique](../critique/SKILL.md) judges work against its brief, including visual quality. Reuse their evidence where relevant. When reviewing for another workflow, return the findings to its owner; the [routing guide](../../references/Routing.md) explains the boundaries. ## Set scope and independence Identify the target, purpose, governing requirements, project conventions and intended coverage. Read enough connected context to understand the behavior and its boundaries. Seek a reviewer independent of the work's author when risk, the task or an applicable gate warrants it, subject to host permissions. If unavailable, state the gap; never call a self-review independent. ## Code change 1. Pin the target. Record the base and head or the exact diff and snapshot; include staged, unstaged, and new files when in scope. Resolve refs and confirm the intended changes are present. 2. Find the governing intent in the request, issue, plan, or documented behavior, and find the relevant project standards. Review spec fit and standards as separate passes so neither obscures the other. When no written spec is available, state the limit and continue with evidenced behavior and standards; label open questions about intent. 3. Trace connected callers, contracts, failure paths, and checks to assess correctness and risk. Confirm each finding's triggering condition and practical consequence. Distinguish observed defects and supported risks from preferences or questions. Use existing evidence and focused non-destructive checks when useful. If delegation is unavailable or unnecessary, perform the passes inline; do not describe self-review as independent. 4. When corrections are authorized, follow the [review loop](../../references/ReviewLoop.md), keep changes within scope and recheck material repairs. ## Existing module, system or repository For an explicitly scoped audit, identify entry points, important flows, data and trust boundaries, dependencies and existing checks. Trace connected behavior and failure paths; inspect correctness, security, reliability, integration and other risks relevant to the target. Choose coverage based on the repository's size and structure: read the relevant repository in full when manageable; for a larger one, select a defensible scope using risk and connected behavior. State the areas examined and excluded, why they were selected, and any limits. Reading every file does not prove every execution path is safe. For production-facing targets, follow the relevant contracts through [Secure](../secure/SKILL.md), [Migrate](../migrate/SKILL.md), [Release](../release/SKILL.md), [Deploy](../deploy/SKILL.md) or [Operate](../operate/SKILL.md). Reuse their evidence for threats, preserved state, artifact identity, rollout and recovery; load only the guidance needed for the audit. A review remains read-only even when a supporting skill describes execution. ## Draft plan Review the fixed draft and criteria provided by the requester or [Plan](../plan/SKILL.md). Check outcome fit, evidence, assumptions, authorization and product decisions, dependencies, failure modes and verification. Identify missing decisions and unsupported claims. Return findings and open questions without editing the draft. Plan owns review coordination, repairs and the record of authorization. ## Completion Report confirmed defects, supported but unverified risks, and concrete material improvements as distinct categories. Prioritize by impact and likelihood; give the location, triggering condition, evidence, consequence and useful next step. For an unverified risk, identify the missing evidence. Keep preferences and style suggestions out unless their impact is material. State the target or revision, coverage and exclusions, independence, and actual checks. If no material findings exist, say so without claiming complete safety. Stop after the report unless fixes were already authorized.
Pliki w paczce (2)
- SKILL.md 4 KB
- agents/openai.yaml 1 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · nazmirket/skills | 90 | MIT |
| marcel-bich/marcel-bich-claude-marketplace | 100 | MIT |
| jeremylongshore/tons-of-skills-marketplace | 100 | MIT |
| friendlyinternet/nuxt-crouton | 100 | MIT |
| userKael/rastro | 100 | MIT |
| Archont561/qgis-agent-harness | 100 | MIT |
| lilly-logic-labs/lll-skills | 100 | MIT |
| Come-In-Labs/agent-skills | 100 | MIT |
| minanyang/10x-agent | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 15e1f6d3efd9 (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
- 80Code review
Przegląda zmianę lub moduł pod kątem poprawności, bezpieczeństwa, regresji i brakujących testów przed wdrożeniem. - 90reviews
Przegląda zmianę w kodzie lub pull request pod kątem kondycji kodu, np. na prośbę o code review. - 90code-review
Przegląda pull request według blokujących kryteriów repozytorium i zgłasza tylko ustalenia wskazujące konkretny tryb awarii. - 90change-review
Przegląda gałąź, PR lub zmianę w repozytorium pod kątem standardów i uzgodnionej specyfikacji, raportując obie osie osobno. - 90understand-explain
Wyjaśnia szczegółowo wskazany plik, funkcję lub moduł w bazie kodu. - 90review-pr
Przegląda pull request z GitHuba wskazany numerem lub adresem URL.