AGENTS.md Generator
Build a complete AGENTS.md or CLAUDE.md from your project type. Commands, code style, boundaries, and workflow, pre-filled for your stack.
Free, no signup. Updates live as you type.
# AGENTS.md ## Project Untitled project — a web app. ## Commands - Install deps: `npm install` ## Code style - Match the existing formatting and naming conventions in the file you're editing before introducing your own. - Prefer editing existing files over creating new ones. - No unused imports, no dead code, no commented-out blocks left behind. - Write self-documenting code — comment only the non-obvious "why", never the "what". ## Boundaries - Don't modify files outside the scope of the current task without asking. - Don't add new dependencies without checking they're actually needed. - Don't change CI/CD config or deployment settings unless explicitly asked. ## Testing and git workflow - Run `the test suite` before considering a change complete. - Run `the linter` and fix any new warnings introduced by your change. - Create small, reviewable commits with clear messages describing the "why", not just the "what". - Never force-push, never skip commit hooks, never amend a commit that's already been pushed.
Why AI coding tools need this file
Claude Code, Cursor, and similar tools read AGENTS.md automatically before touching your codebase. Without it, they infer conventions from whatever files they happen to open first — which means inconsistent formatting, dependencies you didn't ask for, and commands that don't match how your project actually runs.
A good AGENTS.md is short and specific: the exact commands to run tests and lint, not a description of your architecture. Boundaries matter more than instructions — telling the assistant what NOT to touch prevents more damage than telling it what to do.
Frequently Asked Questions
Everything to know about AGENTS.md and CLAUDE.md.
It's a project-root file that tells AI coding assistants (Claude Code, Cursor, and others) how your project works — commands to run, code style, and boundaries they shouldn't cross. Without it, the assistant guesses, and guesses are how you end up with a random new dependency or a broken lint config.
Our Work

Mizu AI
Shipping an AI-native automation builder from zero to launch in 6 weeks

Mrsam AI
The RTL-first, no-code website builder that closed a $500K seed round

Aprex
From blank canvas to production app: a precision tool for thought in React
Book a Call