What is agents.md?

This repository is a personal collection of global instructions that its author uses to guide AI coding agents such as OpenCode, Claude Code, and Codex. The core of the project is a single file called AGENTS.md, which is loaded automatically at the start of every coding session and acts as a system level guide telling the AI agent how to behave. The instructions cover things like what the agent should verify before finishing a task, what steps it can safely skip, how to write commit messages and pull request descriptions, and when the agent should stop and ask the human for input instead of guessing. The version published here is described as an anonymized copy of the file the author actually keeps on their own machine, with personal or identifying details removed. The README is short and does not go into detail about the specific rules themselves, those live inside the AGENTS.md file rather than the README. The project follows a broader convention called agents.md, which is a shared format some developers use for writing this kind of instruction file so that different AI coding tools can read it in a consistent way. The author explicitly invites others to copy, fork, or reuse any part of the file for their own setups. This is less a piece of software and more a reference document, useful mainly as an example of how one experienced user structures their standing instructions for AI agents. Anyone curious about prompt engineering for coding assistants, or looking for a starting template for their own AGENTS.md file, could read through it for ideas. It is released under the MIT license, so reuse is unrestricted.