performance-engineering_skill

Overview

This skill helps validate system performance under load, locate bottlenecks via profiling, and apply targeted optimizations to improve responsiveness and scalability. It covers load testing (k6, Locust), profiling for CPU/memory/I/O, and practical optimization patterns for backend, database, and frontend. Use it to establish performance SLOs, detect regressions, and plan capacity.

How this skill works

The skill recommends which test type to run (load, stress, soak, spike) based on the learning goal and provides quick-start examples for k6 and Locust. It guides profiling choices (py-spy, memory_profiler, pprof, DevTools), shows workflows for finding hot paths, and maps optimizations—caching, query tuning, pagination, and frontend improvements—to measured problems. It also includes CI/CD integration patterns and recommended SLO targets to automate regression detection.

When to use it

• Validate API capacity before launch
• Find maximum capacity and failure modes
• Detect memory leaks or resource exhaustion
• Investigate slow responses or high latency
• Add performance tests to CI/CD for regression detection

Best practices

• Choose test type by the question you need answered (load/stress/soak/spike)
• Profile under realistic load and use sampling profilers in production
• Focus optimizations on top hot-paths that account for most resource use
• Set measurable SLOs and fail CI when thresholds are violated
• Re-profile after each change to verify improvement

Example use cases

• Pre-launch capacity planning with k6 and SLO thresholds
• CI pipeline: run lightweight k6 or Locust tests on pull requests to detect regressions
• Production troubleshooting: use py-spy to capture flamegraphs and find CPU hot spots
• Long-running service: run soak tests to reveal memory leaks and validate connection cleanup
• Frontend optimization: improve Core Web Vitals via image optimization, code splitting, and preloading