Make an AGENT.md

"How standards proliferate" — XKCD #927

But seriously, this time we really do need one standard.

The Problem

Every AI coding assistant has its own configuration approach:

Amp: reads AGENT.md
Augment: looks for .augment-guidelines
Claude: looks for CLAUDE.md
Cline: expects .cline_rules
Codex expects AGENTS.md
Cursor: uses .cursorrules
GitHub Copilot: .github/copilot-instructions.md
Jules: looks for AGENTS.md
Roo Code: expects .roorules
Windsurf: expects .windsurfrules

Developers are creating multiple config files for the same information. This is inefficient and error-prone.

The Solution: AGENT.md

A single, vendor-neutral configuration file that all AI coding agents can understand.

Why AGENT.md?

Vendor-neutral: Not tied to any specific AI service, any AI tool developer can contribute
Descriptive: Clear what the file is for
Markdown: Human-readable and easy to maintain
Standardized: One file, all agents
Domain secured: On X, when you type AGENT.md it will redirect to this site. AGENTS.md is squatted.

AGENT.md 101

What is it?

An AGENT.md is a markdown file that tells AI coding agents how your project works. It contains coding standards, project structure, testing commands, and development practices.

Why should I make it?

It helps AI agents understand your codebase faster and follow your team's conventions. Instead of agents guessing or making assumptions, they get explicit guidance.

Who should make it?

Any developer or team working with AI coding assistants. If you want consistent, high-quality AI-generated code that follows your standards.

When should I make it?

As soon as you start using AI coding agents. It's especially valuable for teams and open-source projects where multiple people work with AI assistants.

Where should I put it?

In the root directory of your project, alongside your README.md. All major AI coding agents should look here first.

Template

AGENT.md Content (editable)

# Agent Configuration

## Project Overview
This is a React TypeScript project using Vite for bundling and testing with Jest.

## Development Commands
- `npm run dev` - Start development server
- `npm run build` - Build for production  
- `npm run test` - Run tests
- `npm run lint` - Run ESLint
- `npm run type-check` - Run TypeScript compiler

## Code Style
- Use TypeScript for all new files
- Follow existing naming conventions (camelCase for variables, PascalCase for components)
- Use functional components with hooks
- Prefer const assertions and strict typing
- Use absolute imports with @ alias for src/

## Testing
- Write tests for all new components in `*.test.tsx` files
- Use React Testing Library for component tests
- Mock external dependencies
- Aim for 80%+ test coverage

## Architecture
- Components in `src/components/`
- Utilities in `src/utils/`
- Types in `src/types/`
- API calls in `src/api/`

## Dependencies
- Prefer built-in browser APIs when possible
- Always check if dependency already exists before adding new ones
- Use exact versions in package.json

Rendered

For AI Companies

Help us standardize! Update your AI agents to look for AGENT.md first, then fall back to your current approach.

Benefits for AI Companies

Better user experience: Users only need one config file
Reduced support burden: Clear, standardized configuration
Faster onboarding: Developers can reuse existing AGENT.md files
Industry leadership: Be part of establishing the standard

Simple Implementation

Make your agent check for files in this order:

AGENT.md (the standard)
Your current format (for backwards compatibility)

Join the Standard

This is a community effort. Help us create a better development experience for everyone using AI coding assistants.

AI Companies: Add AGENT.md support to your agents

Developers: Start using AGENT.md in your projects

Everyone: Spread the word about the standard