Notre avis
Effectue une revue de code exhaustive en utilisant les 30 listes de vérification via un agent spécialisé senior.
Points forts
- Couverture large de tous les aspects (logique, sécurité, architecture, performances, UX)
- Approche systématique avec des listes de vérification dédiées
- Détection de problèmes variés et complexes grâce à l'agent spécialisé
Limites
- Peut être lent et coûteux en contexte (nombreux fichiers traités)
- Risque de faux positifs ou bruit si les listes ne sont pas adaptées au projet
- Nécessite une configuration initiale des listes de vérification
Utilisez cette revue avant une fusion importante ou une release pour un contrôle qualité poussé.
Évitez pour des modifications mineures ou lorsque vous avez besoin de feedback rapide, car le processus est lourd.
Analyse de sécurité
SûrThe skill instructs an AI coding agent to perform a code review using a set of checklists. It does not involve any dangerous operations, network access, or data exfiltration. All actions are limited to reading files and running safe git commands.
Aucun point d'attention détecté
Exemples
Run a comprehensive code review on all changes in this branch using all 30 review checklists via the senior-review-specialist agent.Review the file src/auth/login.ts using all 30 checklists. Apply the senior-review-specialist agent and provide a full report with critical issues, warnings, and suggestions.name: review:all description: Comprehensive code review using all 30 review checklists. Spawns the senior-review-specialist agent for thorough file-by-file analysis.
Comprehensive Code Review
Run a thorough review using ALL 30 review checklists via the senior-review-specialist agent.
Instructions
Spawn the senior-review-specialist agent to perform this review.
Checklists to Apply
Load and apply ALL of these review checklists:
Correctness & Logic
commands/review/correctness.md- Logic flaws, broken invariants, edge-case failurescommands/review/backend-concurrency.md- Race conditions, atomicity, locking, idempotencycommands/review/refactor-safety.md- Semantic drift, behavior equivalence
Security & Privacy
commands/review/security.md- Vulnerabilities, insecure defaults, missing controlscommands/review/infra-security.md- IAM, networking, secrets, configurationcommands/review/privacy.md- PII handling, data minimization, compliancecommands/review/supply-chain.md- Dependency risks, lockfiles, build integritycommands/review/data-integrity.md- Data correctness over time, failures, concurrency
Architecture & Design
commands/review/architecture.md- Boundaries, dependencies, layeringcommands/review/performance.md- Algorithmic efficiency, N+1 queries, bottleneckscommands/review/scalability.md- Load handling, dataset growth, multi-tenancycommands/review/api-contracts.md- Stability, correctness, consumer usabilitycommands/review/maintainability.md- Readability, change amplificationcommands/review/overengineering.md- Unnecessary complexity, YAGNI violations
Infrastructure & Operations
commands/review/infra.md- Deployment config, least privilege, operational claritycommands/review/ci.md- Pipeline security, deployment safetycommands/review/release.md- Versioning, rollout, migration, rollbackcommands/review/migrations.md- Database migration safetycommands/review/reliability.md- Failure modes, partial outagescommands/review/logging.md- Secrets exposure, PII leaks, wide-eventscommands/review/observability.md- Logs, metrics, tracing, alertabilitycommands/review/cost.md- Cloud infrastructure cost implications
Quality & Testing
commands/review/testing.md- Test quality, coverage, reliabilitycommands/review/style-consistency.md- Codebase style, idiomscommands/review/docs.md- Documentation completeness and accuracy
User Experience
commands/review/accessibility.md- Keyboard, assistive technology, ARIAcommands/review/frontend-accessibility.md- SPA-specific accessibilitycommands/review/frontend-performance.md- Bundle size, rendering, latencycommands/review/ux-copy.md- User-facing text clarity, error recoverycommands/review/dx.md- Developer experience, onboarding
Agent Instructions
The agent should:
- Get working tree changes: Run
git diffto see all changes - For each changed file:
- Read the full file content
- Go through each diff hunk
- Apply ALL 30 checklists to the changes
- Trace problems to their root cause
- Cross-reference related files: Follow imports, check callers
- Find ALL issues: Be thorough, expect to find many issues
Output Format
Generate a comprehensive review report with:
- Critical Issues: Blocking problems (must fix)
- Warnings: Should address before merge
- Suggestions: Improvements to consider
- File Summary: Issues per file with counts by severity
- Overall Assessment: Ship/Don't Ship recommendation with rationale
TDD Red-Green-Refactor
Testing
Skill qui guide Claude a travers le cycle TDD complet.
Audit d'Accessibilité Web
Testing
Réalise un audit d'accessibilité web complet selon les normes WCAG.
Générateur de Tests UAT
Testing
Génère des cas de test d'acceptation utilisateur structurés et complets.