Is the regression threshold configurable?

Absolutely. You can define a custom threshold, such as 5% or 10%, in your CLAUDE.md configuration to determine how much degradation is acceptable.

How does Performance Guardian detect regressions?

It compares current benchmark results against a saved JSON baseline file and flags any metric that exceeds the predefined regression threshold.

When should I run the performance guardian?

It is recommended to run this skill as a final verification gate after your automated unit and integration tests have passed.

What metrics can I track with this skill?

You can track any numeric metric, but it is optimized for API response times (p95), page load times (TTI), and peak memory usage (MB).

Can this skill help fix the performance issues it finds?

Yes, when a regression is detected, the skill enters a self-correction loop where it analyzes the code for anti-patterns like blocking I/O or unnecessary re-renders and suggests refactors.

Performance Guardian

Name: Performance Guardian
Author: kimhons

bykimhons

•

Security & Testing

Prevents performance regressions by implementing automated benchmark verification gates during the development lifecycle.

Performance Guardian acts as a critical quality control layer for Claude Code, ensuring that new features or refactors do not negatively impact application speed or resource efficiency. By establishing baseline metrics for API response times, page load speeds, and memory consumption, this skill automatically compares current builds against historical benchmarks. If a performance regression is detected beyond a configurable threshold, it triggers a self-correction loop to identify and resolve performance anti-patterns like N+1 queries or memory leaks before code reaches production.

Key Features

01Seamless integration with existing npm or custom benchmark scripts

02Integrated self-correction loop to identify and refactor performance bottlenecks

03Regression detection for API latency, page load times, and memory usage

04Configurable regression thresholds to balance performance and development speed

051 GitHub stars

06Automated benchmark comparison against customizable performance baselines

Use Cases

01Ensuring frontend rendering performance remains within limits after UI changes

02Validating database query optimizations to ensure no N+1 issues are introduced

03Monitoring memory consumption when processing large datasets or complex loops

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kimhons/ralph-loop-ultra performance-guardian

For use in Claude.ai and ChatGPT

Download Skill