jeltz
CLAUDE.md4 starsChanged 9 months agoMITView on GitHub
Every project must provide a Makefile exposing the following standard targets: - make typecheck Runs type checking (e.g., mypy). - make test Runs the test suite (e.g., pytest) with coverage enforcement. - make verify Runs lint + typecheck + tests as a unified verification step. These conventions apply to all code generated or modified by Claude: All dependencies — direct or transitive — must use OSI‑approved open‑source licenses. If a non‑preferred license is used, Claude must justify why no permissive…
In this repository5 other files it gives its agents.
- project-bootstrapSkill
- reviewer-responseSkill
- security-auditSkill
- skeptical-reviewerSkill
- tdd-phase-loopSkill
The file
Show its rules and text
We split this CLAUDE.md into the rules it gives an agent
15 rules. About 0 are common, 12 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 3
Every project must provide a Makefile exposing the following standard targets:
- Distinctive: line 19
If pre‑commit or post‑tool hooks are used, they must call the same commands above.
- Very specific: line 34
Use the standard logging module
- Distinctive: line 48
All dependencies — direct or transitive — must use OSI‑approved open‑source licenses.
- Distinctive: line 65
If a non‑preferred license is used, Claude must justify why no permissive alternative exists.
- Very specific: line 75
Test files must live under tests/
- Distinctive: line 76
Test names must follow:
- Distinctive: line 91
Use Conventional Commits, for example:
- Distinctive: line 113
When Claude is generating or modifying code:
- Distinctive: line 116
Do not write implementation before tests
- Distinctive: line 117
Do not skip refactoring
- Distinctive: line 118
Do not introduce dependencies without checking license compatibility
- Distinctive: line 119
Do not modify unrelated code unless explicitly instructed
- Distinctive: line 120
Do not commit — only generate commit messages
- Distinctive: line 121
Always ensure the final output is consistent with this document
≈ “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
Build, Test, and Development Commands
Every project must provide a Makefile exposing the following standard targets:
Required Targets
make lintRuns static analysis and formatting checks (e.g.,ruff check,ruff format --check).make typecheckRuns type checking (e.g.,mypy).make testRuns the test suite (e.g.,pytest) with coverage enforcement.make verifyRuns lint + typecheck + tests as a unified verification step.
Hooks
If pre‑commit or post‑tool hooks are used, they must call the same commands above.
Coding Style & Naming Conventions
These conventions apply to all code generated or modified by Claude:
Language & Formatting
- Python 3.11+
- PEP 8 formatting
- Google‑style docstrings for public modules, classes, and functions
- Type hints required everywhere (explicit types for public APIs)
Logging & Error Handling
- Use the standard
loggingmodule - Log to stderr only
- No
printstatements in production code - Fail fast internally; handle exceptions at clear call boundaries
Naming
snake_casefor modules, variables, and functionsPascalCasefor classes- Constants in
UPPER_SNAKE_CASE
Dependency & Licensing Requirements
All dependencies — direct or transitive — must use OSI‑approved open‑source licenses.
Preferred (permissive)
- MIT
- BSD 2‑Clause / 3‑Clause
- Apache 2.0
Allowed (only if necessary)
- GPL v2
- LGPL (any version)
Prohibited
- GPL v3
- AGPL
- SSPL
- Any other restrictive or viral copyleft license not explicitly allowed
If a non‑preferred license is used, Claude must justify why no permissive alternative exists.
Testing Guidelines
General
- TDD (red/green/refactor) is mandatory
- Use
pytest - Enforce 100% coverage unless the project explicitly defines an exception
- Test files must live under
tests/ - Test names must follow:
- Files:
tests/test_*.py - Functions:
test_*
- Files:
Test Quality Expectations
- Tests must assert behavior, not implementation details
- Tests must be deterministic
- Tests must be isolated and not depend on external state
- Fixtures should be used for setup, not ad‑hoc inline scaffolding
Commit & Pull Request Guidelines
Commit Messages
Use Conventional Commits, for example:
feat: add user session managerfix: handle missing policyrefactor: simplify token parsingtest: add coverage for edge cases
Pull Requests
PRs should include:
- A clear summary of changes
- Linked issues (if applicable)
- Evidence that
make verifypassed - Notes on any architectural or behavioral implications
Review Expectations
- Code must materially satisfy the requirements
- Non‑blockers should be noted but do not prevent merging
- Blockers must be resolved before approval
Claude‑Specific Expectations
When Claude is generating or modifying code:
- Follow TDD strictly
- Do not write implementation before tests
- Do not skip refactoring
- Do not introduce dependencies without checking license compatibility
- Do not modify unrelated code unless explicitly instructed
- Do not commit — only generate commit messages
- Always ensure the final output is consistent with this document
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.

