AI Agent Project Starter

CLAUDE.md and AGENTS.md Generator

Turn an app idea into a CLAUDE.md / AGENTS.md file, a clear spec and a step-by-step prompt plan for Claude Code, Codex or Cursor. Answer 4 short questions - no coding needed.

Step 1 of 4Your idea
Step 1 of 4

What do you want to build?

Describe it like you would to a friend. A few sentences is enough.

Tip: say what problem it solves and what a person does with it on a normal day.

Or start from an example:
Nothing is sent anywhere - your answers stay in this browser.

What is CLAUDE.md?

CLAUDE.md is the memory file of Claude Code. Claude Code reads it automatically when it starts in a folder, so it knows your project, your commands and your rules without you repeating them in every chat. Put it in the root of the project. A personal one in ~/.claude/CLAUDE.md holds the rules you want in every project.

CLAUDE.md can import other files with @path. The simplest setup is one line, @AGENTS.md, plus a few notes that only Claude needs.

New to Claude Code? The Claude Code tutorial for beginners covers installing it, plan mode and your first project step by step.

What is AGENTS.md?

AGENTS.md is a plain text file in the root of your project that tells an AI coding agent how to work on it: what the project is, how to run and test it, and the rules to follow. Think of it as a README written for the agent instead of for people. It is an open format read by OpenAI Codex, Cursor, GitHub Copilot, Gemini CLI and many other agents.

The agent reads it at the start of every session, so you do not have to repeat the same instructions in every chat.

AGENTS.md vs CLAUDE.md

CLAUDE.md
  • Read by Claude Code
  • Can load AGENTS.md with one line: @AGENTS.md
  • Keep only Claude-specific notes here
AGENTS.md
  • Read by Codex, Cursor, Copilot, Gemini CLI and more
  • Holds the facts about your project and the rules
  • Write it once, every agent can use it

Best of both: put everything in AGENTS.md and make CLAUDE.md a short file that imports it. That is what this tool creates.

CLAUDE.md example

@AGENTS.md

## Claude Code only
- Use plan mode for each new step and show me the plan first.
- Run npm test before you say a task is done.
- Keep this file short. Project facts go in AGENTS.md.

AGENTS.md example

# AGENTS.md

## Project
Habit Buddy: a web app for tracking daily habits. Spec: SPEC.md

## How to work
- Work on one step at a time, then stop and wait.
- Ask before adding a new package or paid service.
- Never write API keys in code. Use .env (in .gitignore).
- Add a test for each feature and run all tests before finishing.

## Commands
- Install: npm install
- Run: npm run dev
- Test: npm test

CLAUDE.md best practices

  • Keep it short - about one page. It is loaded into every session, so every line costs the agent attention.
  • Start with the commands: how to install, run and test the project.
  • Write rules as clear instructions ("Run the tests before you finish"), not as background stories.
  • Add a rule when the agent makes the same mistake twice, and delete rules that no longer matter.
  • Put project facts in AGENTS.md and import it with @AGENTS.md, so Codex, Cursor and Copilot read the same facts.
  • Point to longer documents such as SPEC.md instead of pasting them in.
  • Never put API keys, passwords or personal data in it - it is usually committed to git.

Claude Code prompts for your first project

This is the prompt plan the generator writes for the habit tracker example. Paste one prompt at a time and check the result before the next one.

  1. 1 Make a plan, no code yetRead SPEC.md and AGENTS.md. Do not write code yet. Propose the simplest stack for this project and explain why in plain words. Then list 5 to 8 small build steps. Ask me any questions you have first.
  2. 2 Set up the projectSet up the project with the stack we agreed. Make it start with one command and show a simple placeholder home screen. Fill in the Commands section of AGENTS.md.
  3. 3 Build the first featureBuild this feature from SPEC.md: "Add, edit and delete habits". Keep it simple. Add a test for it, run the app and all tests, and tell me how to try it.
  4. 4 Check against the specGo through the "Done when" list in SPEC.md. Tell me what works, what is missing and what is broken. Then fix the problems one at a time.
  5. 5 Write the READMEWrite a short README: what the app does, how to install and run it, and how to put it online for free if that is possible.

FAQ

Do I need to know how to code?

No. Choose "Never coded" in step 3 and the files tell the agent to explain each change in plain words and to work in small steps you can check yourself.

Does Cursor read AGENTS.md?

Yes. Put AGENTS.md in the root of the folder you open in Cursor and its agent picks it up.

Does Claude Code read AGENTS.md?

Claude Code reads CLAUDE.md. The CLAUDE.md made here starts with @AGENTS.md, so Claude Code loads both.

What should I put in CLAUDE.md?

What the project is, the commands to install, run and test it, and the rules the agent must follow, such as working in small steps or asking before adding packages. Keep facts every agent needs in AGENTS.md and import it with @AGENTS.md.

How long should AGENTS.md be?

Short - about one page. The agent reads it in every session, so long files waste its attention. Put details in SPEC.md or other docs and link to them.

Is my idea sent anywhere?

No. The files are made in your browser from your answers, and your answers are kept only in this browser.

How much does it cost?

This tool is free. The AI agents have their own plans: Claude Code needs a Claude subscription or API account, Codex works with a ChatGPT plan or API key, and Cursor has a free tier.