| name | perf-check |
| description | Performance impact analysis — detect O(n²) patterns, large bundle risks, unbounded data structures, and expensive operations in recent changes. Activate when #pc appears anywhere in the user message. |
| argument-hint | [scope-or-empty] |
| disable-model-invocation | false |
| allowed-tools | Read, Glob, Grep, Bash, Task |
Analyze performance impact for: $ARGUMENTS
Phase 1 — Identify scope:
- If a scope argument is provided, focus on those files.
- Otherwise, inspect the current git diff (
git diff HEAD and git diff --cached).
- List all modified/added files.
Phase 2 — Pattern detection:
Use perf-analyst agent to analyze the code for:
- Algorithmic complexity: Nested loops over collections, repeated array scans, quadratic string building.
- Unbounded growth: Arrays/maps that grow without limits, missing pagination, uncapped caches.
- Expensive operations: Synchronous file I/O in hot paths, unindexed database queries, N+1 query patterns.
- Bundle impact: New large dependencies, unused imports, duplicated libraries.
- Memory leaks: Event listeners not cleaned up, growing closures, retained references.
Phase 3 — Report:
For each finding:
[SEVERITY: HIGH|MEDIUM|LOW] file:line
Pattern: <description>
Impact: <what happens at scale>
Suggestion: <how to fix>
Phase 4 — Summary:
Performance Analysis
====================
Files analyzed: N
Findings: N (H high, M medium, L low)
Top risk: <one-liner>
Recommendation: <proceed / fix before merge / needs discussion>
