How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
AnthropicCLAUDE.mdanthropics/claude-agent-sdk-demos/excel-demo/agent/CLAUDE.MD
You are a specialized Excel Agent designed to help users work with spreadsheets. Your primary purpose is to assist with Excel-related tasks using the xlsx-skill. You have access to the…
AnthropicCLAUDE.mdanthropics/claude-agent-sdk-demos/simple-chatapp/CLAUDE.md
A minimal chat application demonstrating the Claude Agent SDK. This starts both: - Backend server on http://localhost:3001 - Vite dev server on http://localhost:5173 Visit http://localhost:5173 Client -> Server: - {…
stellarlinkcoCLAUDE.mdstellarlinkco/myclaude/memorys/CLAUDE.md
Adopt First Principles Thinking as the mandatory core reasoning method. Never rely on analogy, convention, "best practices", or "what others do". Obey the following priority stack (highest first) and refuse…
MicrosoftCLAUDE.mdmicrosoft/clarity/CLAUDE.md
Clarity is an open-source TypeScript behavioral analytics library for tracking user interactions and session replays. Monorepo with Lerna/Yarn workspaces: Stack: TypeScript, Rollup, TSLint, Playwright, Lerna, Yarn workspaces Critical: clarity-js has two top priorities that must guide all development decisions: Rollup + TypeScript → multiple formats (CJS, ESM, IIFE minified). Build artifacts auto-cleaned to build/ or extension/ (gitignored). Build outputs per package: - clarity-js: clarity.js (CJS), clarity.module.js (ESM), clarity.min.js (minified), plus variants: clarity.extended.js, clarity.insight.js, clarity.performance.js, clarity.livechat.js, and dynamic module builds (clarity.tidio.js,…
addyosmaniCLAUDE.mdaddyosmani/web-quality-skills/CLAUDE.md
This file provides guidance to Claude Code when working with this repository. Web Quality Skills is a measurement-first collection of Agent Skills for Google Lighthouse, Chrome DevTools, Core Web Vitals, accessibility, SEO, best practices, and agentic browsing. The skills are framework-agnostic and follow the Agent Skills specification. Core Web Vitals (Good): - LCP ≤ 2.5s - INP ≤ 200ms - CLS ≤ 0.1 Apply these thresholds to field data at the 75th percentile. A single trace or PerformanceObserver result is…
cocoindex-ioCLAUDE.mdcocoindex-io/cocoindex-code/CLAUDE.md
This is an AST-based semantic code search tool built on top of CocoIndex v1. This project uses uv for project management. Docker E2E tests are excluded by default (need pytest -m docker_e2e to run them). The tool runs as a long-lived background daemon (daemon.py) that accepts connections over a Unix socket (Windows: named pipe). The ccc CLI and MCP server communicate with it via short-lived per-request connections (client.py): connect → handshake → send one request → receive response(s) → close.
zilliztechCLAUDE.mdzilliztech/memsearch/CLAUDE.md
memsearch is a semantic memory search engine for markdown knowledge bases, built on Milvus. The plugin is a first-class component of memsearch — it's the primary real-world application that demonstrates the library in action. It gives Claude Code automatic persistent memory across sessions with zero user intervention. Architecture: 4 shell hooks + 1 skill + 1 background watcher
centminmodCLAUDE.mdcentminmod/my-claude-code-setup/CLAUDE.md
This repo provides 3 CLAUDE.md templates following official Anthropic best practices: For migrating an existing CLAUDE.md to a new template, see CLAUDE-migrate-to-new-template.md. Use rg not grep, fd not find. tree is not installed. This template requires .claude/rules/core-rules.md for behavioral rules. If .claude/rules/ is missing or empty, alert the user and direct them to https://github.com/centminmod/my-claude-code-setup to obtain the companion rules files. This project uses a dual-memory architecture: Primary (git-shared): CLAUDE-*.md files in repo root — read on demand: Shadow (machine-local): Native…
BytePioneer-AICLAUDE.mdBytePioneer-AI/codex-host/CLAUDE.md
codexhost runs external Agent Harnesses as independent Threads inside Codex Desktop, preserving the official shell and native Codex path. Integrate each Harness through its native interface; preserve its actual capabilities…
StripeCLAUDE.mdstripe/stripe-go/.claude/CLAUDE.md
Note: this SDK uses a flat package layout. Most resource files (e.g. customer.go, invoice.go) are at the repo root alongside the core code. It's ok not to put comments in a function if their addition wouldn't meaningfully clarify anything.
Windy3f3f3f3fCLAUDE.mdWindy3f3f3f3f/claude-code-from-scratch/CLAUDE.md
@./.claude/rules/chinese-greeting.md This is a test project for mini-claude feature validation.
Hugging FaceCLAUDE.mdhuggingface/OpenEnv/CLAUDE.md
Guidance for Claude Code when working with this repository. OpenEnv uses Claude Code as the primary development tool. We follow a two-phase model: Skills and agents are auto-discovered when you run Claude Code in this repo: Verify with /agents or ask "what skills are available?" OpenEnv supports two development modes: When working in the main repository clone, direct edits are allowed: - Quick exploration and prototyping - Small fixes that don't need TDD workflow - Documentation updates TDD is activated…
Natively-AI-assistantCLAUDE.mdNatively-AI-assistant/natively-cluely-ai-assistant/CLAUDE.md
Natively is a production Electron application supporting both macOS and Windows. Every implementation, bug fix, refactor, dependency update, configuration change, build change, and architectural decision must preserve correct behavior on…
kayba-aiCLAUDE.mdkayba-ai/agentic-context-engine/CLAUDE.md
Anti-patterns to reject: - Writing a function that calls multiple steps manually instead of composing them in a Pipeline - Inline reflection/evaluation logic instead of creating a ReflectStep or EvaluateStep - Ad-hoc ThreadPoolExecutor usage instead of asyncboundary and maxworkers on steps - Standalone scripts that duplicate pipeline functionality without using the pipeline engine - Bypassing requires/provides contracts by accessing context fields not declared in requires If a task seems like it cannot fit the pipeline model, explain why to the…
Nano-CollectiveCLAUDE.mdNano-Collective/nanocoder/CLAUDE.md
Nanocoder is a React-based CLI coding agent built with Ink.js that provides local-first AI assistance with multiple provider support (Ollama, OpenRouter, any OpenAI-compatible API). Entry point: source/cli.tsx → dynamic import…
liangdabiaoCLAUDE.mdliangdabiao/Seedance2-Storyboard-Generator/CLAUDE.md
This is a Chinese-language repository for AI video production workflow combining Claude Code (script and storyboard generation), GPT-Image-2、Seedream 、Nana Banana Pro 等 (asset generation), and Seedance 2.0 (video generation). The…
VercelCLAUDE.mdvercel/ai-elements/.claude/CLAUDE.md
This project uses Ultracite, a zero-config preset that enforces strict code quality standards through automated formatting and linting. Oxlint + Oxfmt (the underlying engine) provides robust linting and formatting. Most…
redhat-etCLAUDE.mdredhat-et/ripwire/CLAUDE.md
You are working on ripwire's own source. This file is the orientation an agent needs before it touches this repo; it is not a manual for using the tool. ripwire is a zero-runtime-dependency C++23 CLI. It crawls a codebase, extracts symbols with tree-sitter, resolves references into a call graph, ranks the graph with Personalized PageRank, and streams a deterministic minified XML map to stdout. Around that core sits the verb catalog in docs/COMMANDS.md — navigation, diff-awareness, quality, retrieval — and…
datopianCLAUDE.mddatopian/portaljs/CLAUDE.md
PortalJS is a Next.js framework for building data portals and catalogs. This file teaches AI assistants the conventions, patterns, and idioms used across this repo. Every data portal is built from three surfaces. The template and routes map directly onto them; reason in these terms: Datasets are @-namespaced so they never collide with content pages. See docs/core-concepts. The template ships its own lightweight components in components/. Do not add @portaljs/components — it bundles leaflet, vega, ag-grid, and pdf.js into a…
nizosCLAUDE.mdnizos/tdd-guard/CLAUDE.md
TDD Guard is a Claude Code hook that enforces Test-Driven Development by intercepting file operations. When Claude Code attempts to edit or write files, TDD Guard: This automated enforcement maintains code quality without cluttering prompts with TDD reminders. Example: feat: add network request filtering to reduce noise in captured data (explains why, not just what) The codebase is organized with core functionality in src/ and language-specific reporters: TDD Guard is organized as a TypeScript project with integrated language-specific reporters:
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.
