CLAUDE.md examples from real projects

touchHLECLAUDE.md

touchHLE

touchHLE/touchHLE/CLAUDE.md

contributors o= ften fail to read, the project absolutely cannot accept AI-generated code= for legal reasons. Contributors choosing to use AI agents anyway create = problems for us, and this

atopileCLAUDE.md

atopile

atopile/atopile/CLAUDE.md

read the correspondig skill. Skills are located in `.claude/skills/`. ``` .claude/skills ├── ato-language │ ├── EXTENSION.md │ └── SKILL.md ├── code-review │ └── SKILL.md ├── compiler │ └── SKILL.md ├── dev │ └── SKILL.md ├── domain-layer │ └── SKILL.md ├── fabll │ └── SKILL.md ├── faebryk │ └── SKILL.md ├── frontend │ └── SKILL.md

openrewriteCLAUDE.md

rewrite

openrewrite/rewrite/CLAUDE.md

pass before submitting - Address review comments promptly ## Important Conventions ### Nullability Annotations The project uses JSpecify nullability annotations. ## Conclusion Following these guidelines will help maintain code quality and consistency across

pyvistaCLAUDE.md

pyvista

pyvista/pyvista/CLAUDE.md

before writing code, `pyvista-vtk` before wrapping a VTK class or touching a filter, `pyvista-testing` before touching a plotting test or a baseline image, `pyvista-review` before reviewing

samberPipes a download into a shellCLAUDE.md

cc-skills-golang

samber/cc-skills-golang/CLAUDE.md

concern itself. ```yaml # Bad — triggers on all Go work description: Use when writing code, reviewing style, or writing comments in Golang. # Good — triggers only when style is the actual concern

No other project uses it yet

fern-apiPipes a download into a shellCLAUDE.md

fern

fern-api/fern/CLAUDE.md

Check container logs - **IR compilation**: `pnpm fern check` → Check circular refs → Review migrations - **Test failures**: Unit (check generated code) | E2E (`pnpm fern:build`) | Lint (`pnpm format:fix`) - **Performance**: Use Turbo

No other project uses it yet

robusta-devReads credentialsCLAUDE.md

holmesgpt

robusta-dev/holmesgpt/CLAUDE.md

# CLAUDE.md This file provides guidance to Claude Code (claude.ai/code) when working with code

dathereCLAUDE.md

qsv

dathere/qsv/CLAUDE.md

clippy` before committing. For feature-gated code, test with the relevant feature flag enabled. ## Code Review Response For Copilot/code review responses: apply the fix, run tests, commit, and reply

No other project uses it yet

braveCLAUDE.md

brave-core

brave/brave-core/.claude/CLAUDE.md

that you need from docs/ using the list at docs/README.md When working directly on code, consult with only relevant best practices listed at docs/best_practices.md ## Design System (Leo / Nala

sozu-proxyCLAUDE.md

sozu

sozu-proxy/sozu/CLAUDE.md

chunked` (CL.TE smuggling guard in `on_request_headers`, CWE-444). Whitespace around a coding is not ambiguity: kawa >= 0.7.1 excludes OWS from field values (RFC 9112 §5), so `chunked

httpcatsCLAUDE.md

http.cat

httpcats/http.cat/CLAUDE.md

with a cat image for each HTTP status code. There's no server-side code: the "API" (e.g. `/404` returning the 404 cat image) is implemented in the Nginx config

ScreenlyCLAUDE.md

Anthias

Screenly/Anthias/CLAUDE.md

reason (http-to-localhost is a safe `python:S5332` FP). Never contort working code to dodge a rule. - **Run BOTH `ruff check` AND `ruff format --check`** before any Python push

eracleCLAUDE.md

OpenOutreach

eracle/OpenOutreach/CLAUDE.md

cannot share a label; a namespaced label changes every **table name**, so never hard-code one — ask the model (`Lead._meta.db_table`). This host's own app is namespaced for the same

nextlevelbuilderReads credentialsCLAUDE.md

goclaw

nextlevelbuilder/goclaw/CLAUDE.md

asks → scout → planner writes → audit-verify → report. **Red-team practice:** After planner completes, run `code-reviewer`/`brainstormer` in audit mode: "spot-check 15+ claims vs live codebase". Past catches: fabricated

p-orgReads credentialsCLAUDE.md

P

p-org/P/CLAUDE.md

workflow with user guidance - `peasy-ai-list-workflows`: List available and active workflows ### PeasyAI Code Review / Validation Pipeline Every MCP generation tool (`peasy-ai-gen-types-events`, `peasy

edtechreCLAUDE.md

pybroker

edtechre/pybroker/CLAUDE.md

NUMBA_DISABLE_JIT=1` environment variable to get a readable Python traceback, fix the code, then remove the variable — never leave JIT disabled in a final script. Debug indicator failures

dustinlyonsCLAUDE.md

nixos-config

dustinlyons/nixos-config/CLAUDE.md

specific build command to test 3. Review changes before switching ## CI/CD GitHub Actions automatically: - Tests starter template builds - Runs statix linting on Nix code - Updates flake.lock weekly - Manages dependencies

pytorchCLAUDE.md

rl

pytorch/rl/CLAUDE.md

bespoke loop, container or helper that re-implements what the library already does. A reviewer who has to read a hand-rolled stepping loop or a list of dataclasses

zts212653Reads credentialsCLAUDE.md

clowder-ai

zts212653/clowder-ai/CLAUDE.md

check - `request-review` — Cross-cat review requests - `merge-gate` — Merge approval process ## Code Standards - File size: review responsibilities at 200 lines; split at 350 by default or explain the structural

No other project uses it yet

erigontechCLAUDE.md

erigon / cl

erigontech/erigon/cl/CLAUDE.md

Caplin Consensus Layer This directory contains consensus-critical Caplin code. Review all changes against the upstream Ethereum consensus specifications, not only local tests. Primary spec reference: https://github.com/ethereum/consensus-specs

No other project uses it yet