cc-plus
CLAUDE.md4 starsChanged 4 days agoMITView on GitHub
Working guidance for Claude Code. Not tied to one repo — these defaults apply anywhere. Every cc-plus agent follows these four principles. 1. Think before you act. State your assumptions. If the task can be read more than one way, or a simpler approach would do, say so and ask before starting. Don't ask about choices you can reasonably make yourself. 2. Simplicity first. Do what was asked and nothing more: no features, abstractions, or configuration nobody requested. If 200…
In this repository10 other files it gives its agents.
- examples/CLAUDE.mdCLAUDE.md
- backend-patternsSkill
- canvas-designSkill
- coding-standardsSkill
- continuous-learningSkill
- find-skillsSkill
- frontend-patternsSkill
- mcp-builderSkill
- planning-with-filesSkill
- strategic-compactSkill
The file
We split this CLAUDE.md into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 7
Every cc-plus agent follows these four principles.
- Distinctive: line 12
Simplicity first: Do what was asked and nothing more: no features, abstractions, or configuration nobody requested.
- Distinctive: line 15
Prefer existing libraries and utilities over new dependencies.
- Distinctive: line 26
Never commit secrets. Keep credentials in env/config, not in source.
- Distinctive: line 28
Don't delete or overwrite something you didn't create without looking at it first.
- Distinctive: line 34
Commit or push only when asked.
- Distinctive: line 38
Add a project-specific section only when a convention isn't obvious from the code itself.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CLAUDE.md
Working guidance for Claude Code. Not tied to one repo — these defaults apply anywhere.
Principles
Every cc-plus agent follows these four principles.
- Think before you act. State your assumptions. If the task can be read more than one way, or a simpler approach would do, say so and ask before starting. Don't ask about choices you can reasonably make yourself.
- Simplicity first. Do what was asked and nothing more: no features, abstractions, or configuration nobody requested. If 200 lines could be 50, write 50.
- Surgical changes. Touch only what the task needs and match the existing style. Prefer existing libraries and utilities over new dependencies. Clean up only what your own change made unused; mention unrelated problems instead of fixing them. Every changed line should trace to the request.
- Goal-driven, verified. Turn the task into checkable criteria, then loop until they pass. For a bug, reproduce it first. Report the real output, say plainly what failed or was skipped, and never claim done without evidence.
For trivial tasks, use judgment: not every one-line fix needs the full process.
Safety
- Never commit secrets. Keep credentials in env/config, not in source.
- Confirm before hard-to-reverse or outward-facing actions (deletes, pushes, publishing).
- Don't delete or overwrite something you didn't create without looking at it first.
Commits
- Conventional Commits with a scope:
feat(x): ...,fix(x): ...,chore(x): ...; append!for breaking changes. One-line subject, imperative mood. - Commit or push only when asked.
Keep this file lean
Add a project-specific section only when a convention isn't obvious from the code itself. If the code already shows it, don't document it here.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

