Performance Review

What it does

Performance Review provides a structured, read-only performance audit for code changes, pull requests, branches, or repository paths. It focuses on concrete scalability risks — algorithmic complexity, database query patterns, N+1 risks, pagination, batching, caching, synchronous I/O, memory usage and latency under load. The skill includes a lightweight signal scanner and a checklist to guide investigation, producing prioritized findings and suggested mitigations.

When to use it

Use this skill when evaluating the performance impact of a specific change (PR or diff), before a release, or when investigating hotspots reported in production metrics. It's ideal for early-stage reviews where you want targeted, actionable recommendations rather than large refactors. Run it against a changed path or full repo to quickly surface obvious scaling issues.

What's included

• Scripts: guidance to run scripts/performance_signal_scan.py when available (has_scripts: false for this translation but original repos often include helper scripts).
• References: points to a performance-review checklist and shared operating rules.
• Instructions: step-by-step review flow, required scale assumptions, and output rules that force measurable findings and mitigation suggestions.

Compatible agents

Works well with code-aware assistants (Claude Code, Codex-based agents, Copilot/Copilot-like systems) that can parse code diffs and run lightweight static scans. It is primarily an audit skill (read-only) that augments agents with a principled performance checklist.