[BUG] Claude Code does not read AGENTS.md at conversation start
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Claude Code does not automatically read AGENTS.md at the start of a conversation, even though it's a standard file designed to provide repository-level instructions to agents.
AGENTS.md is ignored unless the user explicitly asks Claude to read it, leading to repeated violations of documented rules (commit format, testing policy, git behavior) across conversations.
What Should Happen?
When a repo contains AGENTS.md, Claude should read it before doing any work.
Error Messages/Logs
Steps to Reproduce
Create an AGENTS.md in the repo root with some instructions about suggesting a commit message, such as "Make it concise and one sentence only, include all uncommitted changes, follow historical commit message pattern", Then make some changes and ask Claude to suggest a commit message, and it will not do what's in AGENTS.md unless explicitly told to look at it.
Claude Model
Opus
Is this a regression?
I don't know
Last Working Version
_No response_
Claude Code Version
2.1.37 (Claude Code)
Platform
Anthropic API
Operating System
Windows
Terminal/Shell
WSL (Windows Subsystem for Linux)
Additional Information
_No response_
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗