AGENTS.md Generator for AI Coding Agents

Describe your repo once and get an AGENTS.md, CLAUDE.md, Cursor rule, .cursorrules or GitHub Copilot instructions file, with stack presets for Node, Python, Go and Rust.

Last updatedHow we build & check our tools
Results will appear here once you enter your values.

How the generator works

Pick a stack preset (TypeScript / Node, Python, Go, Rust or Blank) to fill the tech stack, the five commands and code style. The Python preset, for example, sets uv sync, uv run pytest and uv run ruff check .. Then describe the project, its structure and your rules, one item per line. Leading dashes and numbers are stripped, and a line such as "lib/ - data access" becomes a formatted path entry.

The output updates as you type and only includes sections you filled in: Project overview, Build and test commands, Project structure, Code style, Architecture and conventions, Testing instructions, Do and don't, Commit and PR guidelines, and Security. Choose a format, then copy or download it:

  • AGENTS.md at the repo root.
  • CLAUDE.md, optionally just an @AGENTS.md import plus Claude Code notes.
  • project.mdc for .cursor/rules, with alwaysApply: true frontmatter.
  • .cursorrules, the legacy single-file Cursor format.
  • copilot-instructions.md for the .github folder.

Which file does each agent read?

AGENTS.md is an open format that OpenAI Codex, Google Jules, Cursor, GitHub Copilot, Aider, Zed and many other tools read. In a monorepo you can add nested AGENTS.md files, and the one closest to the file being edited takes precedence.

Claude Code reads CLAUDE.md, not AGENTS.md. Anthropic's documentation recommends a CLAUDE.md that contains the line @AGENTS.md so both tools share one source of truth; tick the import checkbox and the generator writes exactly that. Claude Code also loads CLAUDE.md files from parent directories and suggests keeping each file under about 200 lines.

Cursor project rules live in .cursor/rules as .mdc files with description, globs and alwaysApply frontmatter. GitHub Copilot reads repository-wide instructions from .github/copilot-instructions.md.

Common mistakes

  • Vague commands. "Run the tests" gives an agent nothing to execute. Write the exact command, such as npm test -- -t "name" for a single test, because agents use these lines to verify their own work.
  • Writing a novel. The file is added to the context of every request. A 600-line file costs tokens on each turn and buries the rules that matter. Watch the line and token estimate under the output, which uses roughly four characters per token.
  • Duplicating content. Keeping separate, hand-edited copies in AGENTS.md and CLAUDE.md means they drift apart. Use the import option instead.
  • Saving the .mdc in the wrong place. Cursor ignores a project.mdc at the repo root; it must sit inside .cursor/rules/.
  • Treating rules as enforcement. These files are guidance. Hard limits, like blocking a command, belong in tool settings or hooks.

Tips for a file agents follow

Write rules that can be checked. "Use 2-space indentation" and "All database access goes through lib/db" work better than "write clean code". Put the reason in the rule when it is not obvious, for example "Don't modify migrations that have shipped, they already ran in production".

Use the Don't list for the mistakes you have actually seen, such as editing lockfiles by hand or adding dependencies without asking. Leave fields blank rather than filling them with generic advice; empty fields drop their section entirely.

Commit the file so the whole team and every agent session share it, and update it whenever a reviewer catches something an agent should have known.

Frequently Asked Questions

Common questions about the AGENTS.md Generator for AI Coding Agents

AGENTS.md is a plain Markdown file at the root of a repository that tells AI coding agents how to work on the project: build and test commands, code style, conventions and pull request rules. It has no required fields. It is read by OpenAI Codex, Google Jules, Cursor, GitHub Copilot, Aider, Zed and other tools, so one file can guide several agents.
From the same team

Stop paying per token — route AI requests to your own GPU

Wide Area AI is a local-first AI gateway: repeated requests hit an edge cache, the rest run free on your own hardware, and the cloud is only a failover. OpenAI-compatible endpoint, free tier.

Start routing — free
VRAM CalculatorCan I Run This AI?AI API Cost CalculatorGPU for Model

Sources & References

AGENTS.md: a simple, open format for guiding coding agents

Defines the AGENTS.md format, the tools that read it, and nested-file precedence.
→Source: agents.md

How Claude remembers your project (Claude Code docs)

How Claude Code loads CLAUDE.md files and @path imports.
→Source: Anthropic

Rules (Cursor docs)

Cursor project rules in .cursor/rules (.mdc frontmatter: description, globs, alwaysApply) and legacy .cursorrules.
→Source: Cursor

Adding repository custom instructions for GitHub Copilot

Location and behavior of .github/copilot-instructions.md.
→Source: GitHub

Explore More Tools

Explore these related tools for your next step.

ai

AI Token Cost Calculator

Estimate LLM API cost from requests, input tokens, cached input, output tokens, batch discounts, and editable per-million-token model rates.

Try it now
ai

Sentiment Tone Classifier

Score reviews, replies and posts positive, negative or mixed, sentence by sentence, with a sentiment model that runs in your browser. Free, private.

Try it now
ai

Image Alt Text Generator

Drop an image to prepare concise alt-text notes, accessibility checks, and model cache behavior without uploading the file.

Try it now
ai

Audio Noise Classifier

Select a short audio clip to route common noise categories for podcast QA, device testing, and support diagnostics.

Try it now
ai

Image Upscale Sharpen Preview

Checks an image's file type and size in your browser before you upscale it elsewhere. It does not change the image. No AI model; nothing is uploaded.

Try it now
ai

Prompt Injection Risk Checker

Paste text bound for an AI model and a model in your browser quotes any prompt injection attempts and rates the risk. Free, private, no signup.

Try it now