Instruction Wizard • CLAUDE.md
Create instructions
for your Claude agent.
Build reusable CLAUDE.md instructions for your agent: its role, project facts, working rules, communication style, security expectations, and checks.
CLAUDE.md is a plain-text instruction file for Claude Code. It helps you repeat fewer project rules. This builder asks for facts you know; you can leave unfamiliar fields blank.
CLAUDE.md guides behavior. It does not enforce tool permissions or make an application secure. Use actual access controls and verify the result. Never put passwords, tokens, or personal records in this file.
Create your file
Where to put the file
- Download and review the generated file. If a CLAUDE.md already exists, compare and merge useful rules instead of replacing it blindly.
- Save it in the project’s top-level folder, beside files such as README.md. Check that the name is exactly
CLAUDE.md, notCLAUDE.md.txt. - Using the browser? Add the file to the GitHub repository and branch attached to your Claude Code task, using your normal review process.
- Check that Claude loaded it. In a supported terminal session, use
/contextand inspect the memory files. Ask Claude to summarize the project instructions and compare its answer with your file.
Keep these rules short and specific. Your next task still belongs in the conversation. For stronger application protections, follow the beginner security lessons.
References
- Anthropic. (n.d.-z). How Claude remembers your project. Retrieved September 24, 2026, from https://code.claude.com/docs/en/memory
- Anthropic. (n.d.-ai). Security. Retrieved September 24, 2026, from https://code.claude.com/docs/en/security
Was this chapter useful?
Report an errorVotes are shared only if you enable optional measurement in Privacy & Preferences.