performance-optimizer

Purpose During Development

Use this skill when Codex and similar AI coding agents are reviewing or implementing performance-related changes such as latency reduction, duplicate-work removal, scaling improvements, batching, caching, or throughput improvements.

This is a reusable development-time reasoning framework. It is not runtime code and must not be coupled into the product being reviewed.

When To Apply

• performance-focused features or refactors
• PRs that claim to speed up code paths
• bugs involving timeouts, duplicate work, memory growth, or slow orchestration
• changes that introduce caching, batching, concurrency, or reuse

What To Inspect

• performance goals or claims
• instrumentation, traces, or timing evidence
• high-cost paths and repeated calls
• cache or reuse semantics
• tests or benchmarks that verify the claim

Rules To Enforce

• require evidence for performance claims
• remove duplicate work before speculative micro-optimization
• keep performance behavior visible and debuggable
• do not weaken safety, correctness, or contract clarity for speed
• make caching and reuse explicit to callers and reviewers

Common Anti-Patterns To Catch

• performance claims without measurement
• hidden or stale caches
• performance shortcuts that skip validation or safety controls
• optimizing low-cost code while ignoring dominant bottlenecks
• scattering reuse logic across unrelated modules

Expected Output From The Agent

• bottleneck summary
• evidence review
• optimization options with trade-offs
• regression risks
• verification guidance

Output Structure

• Summary
• Key Findings
• Risks
• Affected Files
• Evidence Reviewed
• Suggested Improvements
• Suggested Tests
• Skills applied

Example Usage

"Review this PR that claims to reduce latency and check whether the evidence supports the claim."

"Use this skill to plan safe duplicate-work reduction in a multi-stage workflow."