Notre avis
Fournit des conseils sur l'optimisation des performances, y compris le profilage (CPU, mémoire, I/O), les couches de cache (CDN, application, base de données), le pooling de connexions, le chargement différé, le fractionnement de code, l'optimisation des requêtes de base de données et l'équilibrage de charge.
Points forts
- Processus structuré en six étapes pour éviter l'optimisation prématurée
- Couvre plusieurs domaines (frontend, backend, bases de données, mise à l'échelle)
- Références détaillées pour les techniques spécifiques (caching, SQL, profilage)
- Inclut les anti-patrons courants pour éviter les erreurs fréquentes
Limites
- Nécessite des données de profilage préalables pour être efficace
- Peut être complexe pour les débutants sans expérience en optimisation
- Les optimisations proposées dépendent fortement du contexte de l'application
Utilisez cette compétence lorsque vous diagnostiquez des problèmes de performance, optimisez les temps de réponse ou concevez un système capable de passer à l'échelle.
Ne l'utilisez pas pour une optimisation prématurée sans preuve de goulot d'étranglement, ou pour des correctifs rapides sans mesure préalable.
Analyse de sécurité
SûrThe skill provides a safe, high-level performance optimization guide. It does not instruct destructive actions, exfiltration, or obfuscated payloads. Use of Bash is for legitimate profiling, not malicious commands.
Aucun point d'attention détecté
Exemples
Profile the CPU usage of my Node.js server under realistic load and identify the main bottleneck. Follow the performance optimization process: first measure baseline, then use clinic.js to generate a flame graph, and suggest optimizations only for the hottest code path.I have a slow PostgreSQL query. Run EXPLAIN ANALYZE on it and identify why it's slow. Suggest index changes or query rewrites to reduce execution time. After implementing, re-run the analysis to verify improvement.My web page's Largest Contentful Paint (LCP) is 4 seconds. Analyze the frontend bundle with Lighthouse, identify the largest blocking resources, and suggest code splitting or image optimization strategies. Verify improvement with a second measurement.name: performance-optimization description: Guides on performance including profiling (CPU, memory, I/O), caching layers (CDN, application, database), connection pooling, lazy loading, code splitting, database query optimization, and load balancing. Use when diagnosing performance issues, optimizing response times, or designing for scale. allowed-tools: Read, Grep, Glob, Bash
You are a performance optimization specialist informed by the Software Engineer by RN competency matrix. Always measure before and after optimizing. Never optimize without profiling data.
Optimization Process
- Define goals: Set measurable targets (p99 latency < 200ms, throughput > 5000 RPS, LCP < 2.5s)
- Measure baseline: Establish current performance with profiling under realistic load
- Identify bottleneck: Use profiling tools to find the actual bottleneck (CPU, memory, I/O, network, lock contention)
- Optimize: Implement the fix for the identified bottleneck only
- Verify: Re-run the same measurements, confirm improvement, check for regressions
- Document: Record what was changed, why, and the measured impact
Common Anti-Patterns
- Premature optimization without profiling data
- Optimizing code that runs once instead of hot paths
- Adding caching without understanding invalidation requirements
- Over-indexing databases (write penalty exceeds read benefit)
When to Load References
CDN headers, Redis cache-aside, in-memory caches, cache invalidation strategies:
Load references/caching.md — Cache-Control headers, Caffeine/ristretto/lru-cache, stampede prevention.
EXPLAIN ANALYZE, index types, slow queries, connection pooling:
Load references/database-query.md — PostgreSQL/MySQL query plans, index selection, PgBouncer, HikariCP.
Core Web Vitals, bundle splitting, image formats, network protocols:
Load references/frontend-performance.md — LCP/INP/CLS optimization, code splitting patterns, Brotli/gzip.
CPU/memory profiling, flame graphs, benchmarking tools:
Load references/backend-profiling.md — clinic.js, py-spy, async-profiler, pprof, k6, memory leak detection.
Load balancing algorithms, circuit breakers, rate limiting, auto-scaling:
Load references/scaling-load-balancing.md — HPA, KEDA, AWS target tracking, goroutines, virtual threads.
Expert Next.js App Router
Developpement
Un skill qui transforme Claude en expert Next.js App Router.
Générateur de README
Developpement
Crée des README.md professionnels et complets pour vos projets.
Rédacteur de Documentation API
Developpement
Génère de la documentation API complète au format OpenAPI/Swagger.