StudioHow to make a Claude Code rule enforced instead of just requested
The author points out that Anthropic's docs describe CLAUDE.md as context, not enforced configuration. Real enforcement should use permissions.deny rules or PreToolUse hooks that run before tool calls and can block them by exiting with code 2. The author also suggests a test: in a disposable repo, protect a dummy file and ask Claude to edit it to verify execution is actually blocked, noting that a check on one editing tool shouldn't be assumed to cover other tools.
The post distinguishes CLAUDE.md guidance from enforced controls like permissions.deny and PreToolUse hooks, and offers a practical test readers can run themselves.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.