Performance Reviewer

Reviews code for performance bottlenecks, algorithmic complexity, and resource usage issues.

1|Updated Apr 25, 2026
One-click install
npx skills add https://github.com/htxryan/claude-code-config-profiles --skill performance-reviewer-htxryan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Performance Reviewer
Source: https://github.com/htxryan/claude-code-config-profiles/tree/main/.agents/skills/compound/agents/performance-reviewer
Command: npx skills add https://github.com/htxryan/claude-code-config-profiles --skill performance-reviewer-htxryan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code reviews often miss performance problems like quadratic algorithms, unnecessary allocations, unbatched I/O, and missing database indexes, which only surface as slowdowns in production. ## Core Features & Use Cases - Hot Path Analysis: Identifies hot paths in changed code and checks algorithmic complexity, flagging O(n^2) patterns where O(n) alternatives exist. - Resource Auditing: Detects unnecessary allocations, unbatched I/O operations, missing database indexes, and resources that are not properly closed or released. - Parallel Subagent Profiling: Spawns opus subagents to profile different modules in parallel when multiple hot paths exist, and coordinates findings with test-coverage and architecture reviewers via SendMessage. - Use Case: During a pull request review, activate this reviewer to classify each finding as BOTTLENECK, CONCERN, or OK so the team can prioritize measurable performance fixes before merging. ## Quick Start Ask the agent to review the changed code in this pull request for performance bottlenecks and report findings using the BOTTLENECK, CONCERN, and OK format.

Frequently Asked Questions about Performance Reviewer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review code for performance bottlenecks?▼

Read the changed code to identify hot paths, then check algorithmic complexity, unnecessary allocations, I/O batching, database indexes, and resource cleanup. This reviewer classifies each finding as BOTTLENECK, CONCERN, or OK.

What performance issues does a code review check for?▼

The review checks for O(n^2) algorithms where O(n) works, unnecessary allocations or copies, unbatched I/O operations, missing database indexes, and resources that are not properly closed or released.

Can performance review run in parallel across multiple modules?▼

Yes. When multiple hot paths exist, the reviewer spawns opus subagents to profile different modules in parallel, then aggregates the findings into a single structured report.

How does the performance reviewer work with other review agents?▼

It operates as an AgentTeam member in the review phase, spawned via TeamCreate. Cross-cutting findings are shared via SendMessage: test gaps go to test-coverage-reviewer and architectural changes go to architecture-reviewer.

What are the limitations of automated performance code review?▼

Static review identifies likely bottlenecks and concerns but cannot measure actual runtime behavior. Measurable issues flagged as BOTTLENECK should still be confirmed with profiling or benchmarks before optimization work begins.