AGENTS.md

An open, plain-Markdown file placed in a repository that gives coding agents project-specific build, test, style and security instructions.

term

AGENTS.md

category

protocols

short_def

An open, plain-Markdown file placed in a repository that gives coding agents project-specific build, test, style and security instructions.

long_def

Conceived as a 'README for agents', AGENTS.md is plain CommonMark with no required schema; agents scan for conventional headings like '## Build & Test' or '## Code Style'. Agents read the nearest file walking up the directory tree, so monorepo subprojects can ship tailored instructions. Originated in the OpenAI ecosystem and now stewarded by the Agentic AI Foundation (AAIF) under the Linux Foundation; adopted by Codex, Cursor, Zed, Jules, Aider and others.

see_also

agents-txt llms-txt mcp

etymology_origin

Open Markdown format published in the OpenAI ecosystem (repository at github.com/openai/agents.md); plain CommonMark with no required schema, now stewarded by the Linux Foundation's Agentic AI Foundation (AAIF).

related_to

agents-txt llms-txt robots-txt mcp

contrast_with

Unlike agents.txt, which declares site-level access rules and capabilities for visiting agents, AGENTS.md instructs a coding agent how to build, test and style a specific codebase — repository guidance, not access policy.

example

AGENTS.md is supported by OpenAI Codex, Cursor, Zed, Google Jules and Aider, with agents reading the nearest file up the directory tree in a monorepo.

source

https://agents.md/

status

active

why_it_matters

AGENTS.md is the lowest-effort way to make a repository agent-ready for coding agents — a single Markdown file that every major coding agent now reads.

sameAs

—

bridge_entity

protocols/discovery/agents-md

last_verified

2026-06-15

md_twin

/glossary/agents-md.md

← all The Agentic Web Lexicon · .md · JSON