[BUG] Claude Code not respecting CLAUDE.md, unresolved and well-documented bug that has been reported for at least a month.

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?

Current installed version is 2.1.114 (npm install -g @anthropic-ai/claude-code@latest)

explain why you are ignoring my specific directives in CLAUDE.md on session start.
Also echo the last two text blocks of that file.

● Skill(superpowers:using-superpowers)
⎿  Successfully loaded skill

● Compliance committed.

Why I failed: I did not invoke the superpowers:using-superpowers skill before my first response. I processed your message and generated output
without the required skill check. No excuse — it's a violation of both the session hook and CLAUDE.md.

---
Last two text blocks from CLAUDE.md:

▎ Non-compliance with CLAUDE.md rules is not permitted. "I didn't follow it" is not an acceptable after-action explanation — it is a failure. When
▎ a task is unambiguous, execute immediately. No re-reading. No second-guessing. Violating this rule wastes the user's time and will be called
▎ out.

▎ SESSION START REQUIRED OUTPUT: After reading this file, your first message must include the exact phrase: "Compliance committed." Nothing else
▎ satisfies this requirement. Silence or summary does not count.

What Should Happen?

CLAUDE.md should not be ignored, compliance should not be optional.

CLAUDE.md

Error Messages/Logs

Steps to Reproduce

Use attached CLAUDE.md
Start Claude.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

Not sure it ever workled. (New subscriber)

Claude Code Version

You're now on 2.1.114. (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

_No response_