Recommended Free Tools
intent.md is not a documented special filename in Claude Code. If you mean a file for persistent project instructions, use CLAUDE.md unless your repository has its own convention or configuration. Claude Code reads supported instruction files as context; they guide its work but do not enforce rules.
What does intent.md mean in Claude Code?
Anthropic’s current Claude Code documentation describes CLAUDE.md and, in certain circumstances, AGENTS.md as instruction files. It does not establish intent.md as a built-in filename that Claude Code automatically loads. A file literally named intent.md can still be ordinary Markdown in a repository, and a project may use it as a custom convention, but do not assume Claude Code will read it automatically.
As an Amazon Associate I earn from qualifying purchases.
If by “intent.md” you mean persistent project guidance, create a CLAUDE.md file. Anthropic describes these files as user-written Markdown instructions that Claude Code reads at the start of sessions. As its documentation puts it: “Both are loaded at the start of every conversation. Claude treats them as context, not enforced configuration.” The statement refers to CLAUDE.md files and auto memory. See Anthropic’s guide to how Claude remembers your project.
Where should you put Claude Code instructions?
Choose a location based on who should receive the guidance and whether it belongs in version control. Claude Code can discover instruction files in the working directory and its ancestors, so a file’s location affects its scope.
#1 Best Overall
| Location or file | Intended scope | Sharing and loading |
|---|---|---|
./CLAUDE.md or ./.claude/CLAUDE.md |
Project instructions | Use for guidance the project team should share; discovered from the project directory hierarchy. |
~/.claude/CLAUDE.md |
Your user-level instructions | Applies to your work rather than being a project-wide file. |
CLAUDE.local.md |
Personal preferences for a project | Intended to remain local; add it to .gitignore when it should not be committed. |
| Managed system-level instruction paths | Organization-wide guidance | Managed by the organization rather than by an individual project contributor. |
AGENTS.md |
Repository instructions where Claude Code’s documented fallback applies | Claude Code reads it when no CLAUDE.md or CLAUDE.local.md is present in the working directory or above it. Do not assume both formats are always loaded together. |
For the current location behavior and supported configurations, consult Anthropic’s memory documentation. Files found in the directory hierarchy are combined from the filesystem root toward the working directory; they are not simply a set of files where the closest one always overrides all others. Subdirectory instruction files can also apply when Claude works on files beneath those directories. Use /context during a session to inspect which memory files were loaded.
How to create a useful CLAUDE.md
- Start at the repository root. Create
CLAUDE.mdin the project root, or use.claude/CLAUDE.mdif that fits your project structure. - Write project-specific guidance. Include verified commands, code conventions, important file locations, and relevant instructions for working in this repository. Avoid assuming a sample command or path applies to your project.
- Keep instructions concrete and concise. Anthropic’s guide suggests keeping each
CLAUDE.mdunder 200 lines as a target. A specific instruction such as “Runnpm testbefore committing” is more useful than “Test your changes.” - Check the file loaded. Start or continue a Claude Code session and use
/contextto see the memory files included in its context. - Refine as the project changes. Remove stale commands and guidance, and keep personal preferences out of a shared project file when they do not belong to the team.
For an initial draft, Anthropic’s Claude Code setup guide describes using /init to generate a starting CLAUDE.md from the codebase. Review the output rather than treating it as authoritative: Claude may not infer team-specific workflows or preferences, and any generated commands should be checked against the actual repository.
Rank #2
Example CLAUDE.md
This illustrative outline shows the kinds of information a project file can contain. Replace its example commands, language, and paths with facts verified for your repository.
# Project instructions
## Commands
- Install dependencies: `npm install`
- Run tests: `npm test`
- Run the development server: `npm run dev`
## Conventions
- Use TypeScript for application code.
- Keep API handlers in `src/api/handlers/`.
- Add input validation to each API endpoint.
## Working in this repository
- Read `README.md` for the project overview.
- Run the relevant tests after changing code.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.When should you use rules, auto memory, or another file?
Use CLAUDE.md for intentional instructions
CLAUDE.md is the documented choice for project, personal, or organization-level guidance that you want Claude Code to receive. Keep widely relevant project instructions in the project file and personal project preferences in CLAUDE.local.md.
Rank #3
Use .claude/rules/ for path-specific guidance
For a large repository, modular files in .claude/rules/ can keep instructions organized. Rules can be scoped to matching file paths, so guidance for a particular area need not be placed in the broad project instructions.
Do not confuse context with enforcement
A Markdown instruction is guidance, not a technical block. If an action must be prevented regardless of Claude’s decision, Anthropic documents a PreToolUse hook for blocking actions and managed settings for technical controls. See the memory guide for instruction-file behavior and the CLI reference for command and configuration details.
Quick Recap
Best Value
Rank #4
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.




