My collection of standards, skills, and rules for AI coding agents. (WIP)

I work with Claude Code, Codex Cursor, and whatever else ships next week. They're all decent at writing code. They're all terrible at writing code the way I want it.

This repo is my attempt to fix that — a single place for engineering standards, testing guidelines, and reusable skills that I can point any agent at.

• AGENTS.md — Core engineering standards. Language-agnostic rules for how I want code written, tested, and organized.

More coming: platform-specific standards, reusable skills, prompt templates.

Drop the relevant files into your project or ~/.codex or ~/.claude folder.

Or just steal what's useful and make it your own.

These aren't meant to be comprehensive or "correct" They're meant to be useful. Agents interpret literally, so the rules are explicit. They also love to skip steps, so the workflows are numbered.

If something doesn't work, I change it. If something's missing, I add it.

This is personal, but if you've got ideas or improvements, open an issue. I'm not precious about it.

Built by Daniel Bernal while procrastinating on FlowDeck.