Skip to content
shipanysaas
Get the kit

Free tool

AGENTS.md generator.

Describe your project once and get an AGENTS.md your coding agents can all read, plus the extra file some of them need.

No sign-up. It runs in your browser, so nothing you type is sent or saved.

Describe your project

Skip to your AGENTS.md
Project
Commands

Filled in from your stack. Change any to match your project, or clear one to leave it out.

Rules

One rule per line. Clear a box to leave that section out.

Rules you could check in a code review.

How every change gets tested.

How changes are committed and reviewed.

What your agent must never do, or must ask you about first.

Agents you work with

Each one gets a note on how it finds AGENTS.md, and the extra file if it needs one.

Your AGENTS.md

Save it as AGENTS.md at the root of your repository.

# AGENTS.md

## Project overview

Stack: Next.js. Package manager: npm.

## Setup and commands

- Install dependencies: `npm install`
- Start the dev server: `npm run dev`
- Build: `npm run build`
- Run the tests: `npm test`
- Lint: `npm run lint`
- Type check: `npx tsc --noEmit`

## Code style

- Use TypeScript. Avoid `any`; use `unknown` and narrow it.
- Use Server Components by default. Add `'use client'` only for state, effects or browser APIs.
- Follow the patterns in the surrounding code before adding new ones.

## Testing

- Add or update a test for every change in behaviour.
- Never delete, skip or weaken a failing test to make it pass. Fix the code, or ask.

## Commits and pull requests

- Keep each commit to one change, with a message that says what changed and why.
- In the pull request, describe the change and how you tested it.

## Safety rules

- Never commit secrets, API keys or `.env` files, and never print their values.
- Never add a dependency or change the database schema without asking first.
- Never force-push, rewrite Git history or delete branches.

## Before you finish

1. Run `npx tsc --noEmit`, `npm run lint`, and `npm test`.
2. Fix anything that fails, then run the checks again.
3. Tell me what you changed and anything you could not check.

42 lines · 1.3 KiB

How each agent finds it

  • Claude Code

    Add CLAUDE.md

    Claude Code 2.1.277 and later reads AGENTS.md when your repository has no CLAUDE.md. This import also covers older versions and repositories that keep a CLAUDE.md, and never loads the file twice. Put Claude-only notes below it.

    @AGENTS.md

    Save it as CLAUDE.md in your repository.

    Claude Code docs(opens in a new tab)

  • GitHub Copilot

    Add .github/copilot-instructions.md

    Copilot’s cloud agent, its CLI, code review on GitHub and chat in VS Code read AGENTS.md. Chat on github.com and in Visual Studio, JetBrains IDEs, Eclipse and Xcode reads only this file, so it repeats your key rules.

    # Copilot instructions

Read AGENTS.md at the root of this repository before you change anything. It has the commands, code style, testing and safety rules for this project.

The rules that matter most:

- Never commit secrets, API keys or `.env` files, and never print their values.
- Never add a dependency or change the database schema without asking first.
- Never force-push, rewrite Git history or delete branches.
- Before you finish, run `npx tsc --noEmit`, `npm run lint`, and `npm test`.

    Save it as .github/copilot-instructions.md in your repository.

    GitHub Copilot docs(opens in a new tab)

  • OpenAI Codex

    Nothing to add

    Codex reads AGENTS.md before it starts work, from the repository root down to the folder you run it in. It stops adding files at 32 KiB by default.

    OpenAI Codex docs(opens in a new tab)

  • Cursor

    Nothing to add

    Cursor reads AGENTS.md at the project root and in subfolders. You only need .cursor/rules for rules that apply to certain files.

    Cursor docs(opens in a new tab)

What is an AGENTS.md file?

AGENTS.md is a Markdown file at the root of your repository that tells coding agents how to work on your project: the commands to run, the rules your code follows and what to leave alone. Think of it as a README for agents.

It’s an open format, looked after by the Agentic AI Foundation under the Linux Foundation, so one file serves Codex, Cursor, Claude Code, Copilot and most other agents. There are no required headings: the agent reads whatever you write.

Where to put the files

  • Save AGENTS.md at the root of your repository, next to the README, and commit it so every agent and teammate works from the same rules.
  • In a monorepo, add an AGENTS.md to each package that has its own commands or rules. The file nearest the code being changed takes precedence.
  • Put each extra file at the path on its card: CLAUDE.md at the root, then .github/copilot-instructions.md, .gemini/settings.json or .aider.conf.yml.

How to keep it short and useful

  • Write commands exactly as you type them. “Run the tests” makes the agent guess; “npm test” doesn’t.
  • Write rules you could check in a code review: “store money as integer cents”, not “be careful with money”.
  • Stay well under 200 lines. Add a line when you correct your agent twice for the same thing, and delete lines the day they stop being true.
Read the full AGENTS.md guide

Questions about AGENTS.md.

Is this AGENTS.md generator free?

Yes, and there’s no sign-up. It runs in your browser, so nothing you type is sent to a server or saved.

Which coding agents read AGENTS.md?

OpenAI Codex, Cursor, Google Antigravity, Devin Desktop (Windsurf), JetBrains Junie, Cline, Kiro, Devin, Grok Build, opencode, and Zed read it by themselves. Claude Code does too when your repository has no CLAUDE.md. GitHub Copilot reads it in its cloud agent, its CLI, code review and VS Code chat, but chat on github.com and in other IDEs needs .github/copilot-instructions.md. Gemini CLI and Aider need one setting, which the generator writes for you.

Do I still need a CLAUDE.md file?

Not with Claude Code 2.1.277 or later, as long as your repository has no CLAUDE.md. If you keep one, put @AGENTS.md on its first line so Claude reads both. The import never loads AGENTS.md twice, so it’s safe to leave in.

Can a project have more than one AGENTS.md?

Yes. In a monorepo, put one in each package that has its own commands or rules. The file nearest the code being changed takes precedence, and what you type in chat overrides every file.

How long should an AGENTS.md be?

As short as it can be while it still lists your commands and rules. Claude Code’s docs suggest under 200 lines, Devin reads the first 16 KiB of each file, and Codex stops at 32 KiB by default. Move procedures you need only now and then, like cutting a release, into a skill or a doc.

Want a whole SaaS set up like this?

shipanysaas is a Next.js, Expo and Supabase starter kit with login, teams and billing already built. It ships AGENTS.md files across the repo, up to 49 ready-made skills and 17 written fixes for errors agents run into.

Get the kitTry the live demo

From $49 once, no subscription