AGENTS.md - Un formato simple y abierto para guiar agentes de codificación

AGENTS.md is a simple, open format designed to guide coding agents, utilized by over 20,000 open-source projects. It serves as a dedicated space for providing context and instructions that help AI coding agents effectively work on projects. This format complements traditional README.md files by focusing on the specific needs of coding agents.

1. Setup Commands

   • Install dependencies: pnpm install
   • Start the development server: pnpm dev
   • Run tests: pnpm test

2. Code Style Guidelines

   • TypeScript strict mode
   • Single quotes, no semicolons
   • Preference for functional programming patterns

3. Agent Compatibility

   AGENTS.md works seamlessly with various AI coding agents, including:

   Agent Name	Description
   Codex	General-purpose CLI tooling for AI coding agents.
   Amp	AI coding agent from Google.
   Jules	Another AI coding agent from Google.
   Cursor	AI coding tool for enhanced productivity.
   Factory	Tool for building and managing coding agents.
   Roo	AI coding assistant for various programming tasks.
   CodeAider	AI tool for code assistance and generation.
   Gemini CLI	CLI tool from Google for coding tasks.
   Kilo Code	AI coding agent for streamlined development.
   OpenCode	Open-source coding agent for collaborative projects.
   Phoenix	AI tool for enhancing coding efficiency.
   Zed	AI coding assistant for various programming languages.

4. Nested AGENTS.md Files

   For large monorepos, nested AGENTS.md files can be used within subprojects, allowing tailored instructions for each package.

1. Create an AGENTS.md file at the root of your repository.
2. Include sections that assist agents in working effectively, such as project overview, build commands, and testing instructions.
3. Add any additional instructions relevant to new contributors, such as commit messages and security considerations.
4. For large projects, utilize nested AGENTS.md files for specific subprojects to provide localized guidance.

Are there required fields in AGENTS.md?

No, AGENTS.md is flexible and can be structured using any headings as needed. The agent will parse the text provided.

What happens if instructions conflict?

The closest AGENTS.md file to the edited file takes precedence, while explicit user prompts will override any conflicting instructions.

Will the agent run testing commands automatically?

Yes, if testing commands are listed in AGENTS.md, the agent will attempt to execute them and fix any failures.

Can AGENTS.md be updated later?

Absolutely, AGENTS.md should be treated as living documentation that can be modified as needed.

Price data is not available yet; please visit the official website for more information.

1. AGENTS.md effectively provides a structured format for coding agents, enhancing their ability to understand project requirements and execute tasks efficiently.
2. The separation from README.md files helps maintain clarity for human contributors while offering detailed guidance for AI agents.
3. However, the adoption of AGENTS.md may require a cultural shift within teams that are accustomed to traditional documentation practices.
4. Overall, AGENTS.md is a valuable tool for developers looking to streamline their workflows with AI coding agents, but its success depends on consistent usage and updates.