performance-optimization
A measure-first optimization workflow for fixing proven bottlenecks without premature complexity.
30-Second Summary
This skill tells the agent to measure before optimizing.
It establishes a baseline, identifies the actual bottleneck, fixes the specific issue, measures again, and adds a guard.
Use it when performance is part of the requirement or when monitoring and runtime evidence point to a real regression.
1-Minute Read
What it is
A SKILL.md workflow for performance diagnosis and optimization across frontend and backend surfaces. It emphasizes synthetic and real-user measurements, bottleneck classification, targeted fixes, and regression guards.
When to use it
Use it for load time budgets, slow interactions, Core Web Vitals problems, large data sets, high-traffic paths, or suspected regressions. It helps keep optimization tied to evidence rather than intuition.
How to test it first
Pick one slow route or interaction and ask the agent for a baseline measurement plus the suspected bottleneck class. Do not accept a fix proposal until the measurement points to the work.
Watch out
Premature optimization can add complexity without improving user experience. Require before-and-after numbers and avoid spreading memoization, caching, or bundle changes without a measured reason.