Performance Optimization

Use this skill when performance is an explicit goal or an obvious risk: latency, throughput, memory, CPU, startup time, build time, I/O, database load, concurrency, or scalability. The objective is to make code faster or lighter without changing observable behavior, and to verify improvements with measurements whenever measurement is feasible.

Performance is a first-class design constraint, but correctness, security, and maintainability still set the boundaries. Prefer efficient choices when they are equally clear. Reserve invasive rewrites, clever micro-optimizations, and added complexity for measured hot paths or high-risk workloads.

Activation Criteria

Use this skill when the task includes:

• Explicit performance terms such as optimize, speed up, slow, bottleneck, latency, throughput, memory, allocation, CPU, hot path, scalability, benchmark, profile, N+1, cache, or concurrency.
• Code that runs frequently, handles large data, sits in request/build/render loops, or performs repeated I/O or database work.
• Review of an implementation where algorithmic complexity, excessive allocations, blocking I/O, unbounded concurrency, or resource leaks are likely to cause real cost.
• Choosing between otherwise similar implementations where one is clearly more efficient with no meaningful readability or correctness penalty.

Do not use this skill as the primary lens when: