A drop-in AGENTS.md file that makes coding agents behave like senior engineers, eliminating sycophancy and forcing verification.
agents-md is easy to set up with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
Repository setup guidance
gh repo clone FerroxLabs/agents-mdOpen or clone the template repository.
Move the relevant AGENTS.md content into your project root.
Update project-specific commands, environment rules, and sensitive areas.
This is a single file you drop into any repository. It contains instructions that make AI coding agents (like Claude Code, Codex, Cursor) act more like experienced engineers. They stop blindly agreeing, avoid unnecessary refactors, and always verify their work before claiming completion. It combines best practices from Andrej Karpathy and Boris Cherny.
agents-md is a behavioral scaffold for AI coding agents. It is a single AGENTS.md file that, when placed in a project root, is automatically read by many popular coding agents (Claude Code, Codex, Cursor, Gemini CLI, Aider, Windsurf, Copilot, Devin, and others). The file contains 11 sections: sections 0-9 define the agent's behavior (e.g., push back when the user is wrong, prefer minimal diffs, write verification first, surface ambiguities), section 10 is for project-specific context (stack, commands, layout) that you fill once, and section 11 is a 'Project Learnings' section that starts empty and grows as the agent learns from mistakes. The design is tight (~200 lines) to ensure rules stay loaded. It kills sycophancy, stops drive-by refactors, and forces verification loops. The project is open source under MIT license, has 565 stars, and is actively maintained by Ferrox Labs.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 57 days ago.
565 GitHub stars indicate community interest.
0 open issues signal maintenance load.
MIT license detected.
Improve code quality by ensuring agents verify their changes before reporting done.
Reduce unnecessary refactoring and keep diffs minimal.
Onboard new agents to a project with consistent behavior rules.
Accumulate project-specific learnings over time to avoid repeated mistakes.
Standardize agent behavior across multiple developers using different AI tools.
The file itself is safe, but agents may follow instructions that could modify your codebase. Always review changes before accepting.
If you modify sections 0-9, you may inadvertently reduce the effectiveness or introduce unexpected behavior.
565
Stars
69
Forks
0
Issues
MIT
License
AGENTS.md rules and skills for AI coding agents, distilled from classic software engineering books.
Convert and sync AI coding-agent rule files between formats with zero dependencies.
Generate AGENTS.md from your codebase in one command. Free, instant, no API key.
A drop-in AGENTS.md file that makes coding agents behave like senior engineers, eliminating sycophancy and forcing verification.
Senior software engineers who use AI coding assistants and want more reliable output., Tech leads managing teams that rely on AI agents for code generation., Open source maintainers who want consistent contributions from AI agents.
The setup section provides repository-level starting guidance, not a guarantee of an independently verified installation. Check the official README and release notes.
The file itself is safe, but agents may follow instructions that could modify your codebase. Always review changes before accepting. If you modify sections 0-9, you may inadvertently reduce the effectiveness or introduce unexpected behavior.
Similar or complementary options to evaluate include agent-rules-books, agentsync, agentseed.
Jun 7, 2026
2 security/trust notes recorded.
Setup difficulty is 1/5.