Launch offer: use code LAUNCH20 for 20% off any single pack. See the packs

Free tool

Free AGENTS.md generator

Pick your stack, check the commands, and get a starting AGENTS.md plus a matching CLAUDE.md, Cursor rule and Copilot instructions file. Runs in your browser; nothing is sent anywhere.

Generated files are a starting point written from common defaults. Check every command and rule against your own project. Jump to the result

This tool needs JavaScript. Without it, see the guide to writing an AGENTS.md and copy a template from the free library.

How to use it

  1. Choose your stack. The commands, layout and conventions fill in with common defaults.
  2. Edit anything that does not match your project. Delete lines that do not apply.
  3. Switch tabs for the other files, then copy or download each one into your repository.
  4. Run your agent on a small task and see what it gets wrong. Add one line per repeated mistake.

Which file goes where

FileLocationRead by
AGENTS.mdrepo root (or a subfolder)Cursor, GitHub Copilot, Codex and many others
CLAUDE.mdrepo root, next to AGENTS.mdClaude Code
project-baseline.mdc.cursor/rules/Cursor
copilot-instructions.md.github/GitHub Copilot

Details and sources are in our guide AGENTS.md vs CLAUDE.md vs Cursor rules, and the tips for good lines are in how to write an AGENTS.md.

AI Coding Agent Config Kit

Instruction-file templates for 12 stacks and 12 prompts that write and audit them. 34-page PDF, $14.

Buy on GumroadSee what's insideFree sample templates

FAQ

Questions about the generator

What is an AGENTS.md file?

A plain Markdown file at the root of your repository that tells AI coding agents how to run, test and change your project. It has no required fields. Tools such as Cursor, GitHub Copilot and OpenAI Codex read it, and the closest AGENTS.md to the file being edited wins.

Does Claude Code read AGENTS.md?

Anthropic's docs say Claude Code reads AGENTS.md when there is no CLAUDE.md in the working directory or above it. The safest setup is a CLAUDE.md whose first line is @AGENTS.md, which is exactly what the CLAUDE.md tab here generates.

Is anything sent to a server?

No. The generator runs in your browser. It makes no network calls, uses no cookies or local storage, and has no analytics. Closing the tab discards what you typed.

Can I trust the generated commands?

Treat them as common defaults for the stack you pick. Check every command against your own package.json, Makefile or manifest before you rely on it. The AI Coding Agent Config Kit includes a prompt that makes your agent verify each command.

Will the agent always follow the file?

Not always. These files are guidance, not enforcement. Keep them short and specific, and use permission settings or hooks for anything that must never happen.