An hour of Grill.me on a permissions system. I let the agent work and came back to a half-finished implementation: the context of that conversation had swollen and the original plan no longer fit. I dropped it into a file by phases, one new conversation per phase, and that is when it closed. ---

I have used Cursor for a couple of years. I started on Tab. When I moved to agents, a chunk of my time went to reviewing and fixing what they generated. I built Anti-Cliché.

AI agents speed up how much code we generate. Knip helps find files, exports, and dependencies nobody uses anymore — and turn that hygiene into an automatic project rule.

Why context beats better prompts—and how Matt Pocock's Grill.me skill turned discovery into an explicit part of my AI workflow.

Your website can host installable AI skills — you don't need GitHub. I set this up with Next.js on a production app; the tricky part was getting the www redirect right so npx skills add works, and giving agents persistent product context instead of making things up.

When everyone on the team uses a different AI editor, rules drift between .cursor, .agent, and .claude. We centralized everything in .agents/ with symlinks and tucked the wiring into VS Code so the repo stays focused.

When teams grow, commit styles drift and lint catches up too late. Git Auditor uses versioned shell hooks in .githooks/ to enforce a shared baseline — on your machine, before CI.

PostgreSQL won't auto-update timestamps or clean up related rows — you wire that with triggers. Here's a naming convention that keeps the catalog readable for you and for AI agents on Supabase.

After working on several Supabase-backed projects, I kept hitting the same silent disk problem — log tables that grow forever. Here is the cleanup_logs() function and how to run it safely.