Performance
Per-tool best practices audit across your stack. Finds slow patterns and fixes them without changing business logic.
/performance
Your product works. Now make it fast.
This command spawns specialized agents that audit each layer of your stack against known best practices. The database agent checks your queries, indexes, and RLS policy performance. The connector agent checks your caching strategy, data fetching waterfalls, and bundle size. The designer agent checks your component rendering and image optimization.
The constraint: they fix performance issues without touching your business logic. If a query is slow, they add an index or rewrite the query plan. If a page has a data fetching waterfall, they parallelize the calls. If a component re-renders unnecessarily, they fix the memo boundary. Your features keep working exactly the same way, just faster.
What happens
Each agent reads the best practices documentation for its specific tool — PostgreSQL performance guides, Next.js rendering optimization, Tailwind CSS purging, React re-render prevention. They compare your code against those patterns and flag anything that deviates.
Then they fix what they can. Add missing indexes. Wrap expensive computations in cache directives. Split client bundles. Optimize images. Move data fetching to the server where it belongs.
What you get back
- Performance audit report organized by layer (database, API, rendering, bundle)
- Before/after metrics where measurable
- Applied fixes with explanations of what changed and why
- Lighthouse audit scores for your main pages
Example
/performance
The ideas list page feels slow on first load. The build map page stutters
when marking steps as done. Run a full audit and fix what you can.