How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
AnthropicCLAUDE.mdanthropics/claude-quickstarts/managed-agents/chat-sdk/CLAUDE.md
repo from the inside (`resources: [{type: "github_repository", ...}]`) - **MCP tools**: wire an external data MCP server (`mcp_servers` + `mcp_toolset` on the agent, credentials via a vault, `vault
AnthropicCLAUDE.mdanthropics/claude-quickstarts/managed-agents/mcp-server-typescript/CLAUDE.md
MCP server (TypeScript) × Claude Managed Agents MCP server that exposes Managed Agents session primitives as tools. Two entrypoints sharing one tool set: `src/server.ts` (stdio → Claude Desktop and Claude Code
ag-ui-protocolCLAUDE.mdag-ui-protocol/ag-ui/CLAUDE.md
affected`) instead of using the underlying tooling directly - You have access to the Nx MCP server and its tools, use them to help the user - When answering questions about
open-metadataCLAUDE.mdopen-metadata/OpenMetadata/CLAUDE.md
operator/` — Kubernetes operator - `openmetadata-integration-tests/` — backend API integration tests (`*IT.java`) - `openmetadata-mcp/` — MCP server - `openmetadata-ui-core-components/` — canonical React component library - `openmetadata-ui/src/main/resources/ui/` — React frontend application - `openmetadata
CoplayDevCLAUDE.mdCoplayDev/unity-mcp/CLAUDE.md
Plugin (MCPForUnity/) ↓ Unity Editor API Scene, Assets, Scripts ``` **Two codebases, one system:** - `Server/` - Python MCP server using FastMCP - `MCPForUnity/` - Unity C# Editor package ### Three Layers on the Python Side
Model Context ProtocolCLAUDE.mdmodelcontextprotocol/typescript-sdk/CLAUDE.md
High-Level APIs**: - `Client` (`packages/client/src/client/client.ts`) - Client implementation extending Protocol with typed methods for MCP operations - `Server` (`packages/server/src/server/server.ts`) - Server implementation extending Protocol with request handler registration - `McpServer` (`packages/server/src/server/mcp.ts`) - High-level server
mrexodiaCLAUDE.mdmrexodia/ida-pro-mcp/CLAUDE.md
CLAUDE.md Guidance for working in this repository. ## What this project is IDA Pro MCP Server: exposes IDA Pro / idalib functionality to MCP clients. Main pieces: - `src/ida_pro_mcp/server.py`: MCP server entrypoint - `src/ida_pro_mcp/idalib_server.py
humanlayerCLAUDE.mdhumanlayer/humanlayer/CLAUDE.md
that coordinates approvals and manages Claude Code sessions - `hlyr/` - TypeScript CLI with MCP (Model Context Protocol) server for Claude integration - `humanlayer-wui/` - CodeLayer - Desktop/Web UI (Tauri + React) for graphical approval
AnthropicCLAUDE.mdanthropics/claude-for-legal/CLAUDE.md
commercial-legal, privacy-legal, ...) .claude-plugin/plugin.json # plugin manifest (name, version, description, author) .mcp.json # MCP servers the plugin connects to CLAUDE.md # practice-profile TEMPLATE (see "Plugin CLAUDE.md" below) README.md
AnthropicCLAUDE.mdanthropics/claude-code-action/CLAUDE.md
imports don't need `.js` extensions. - GitHub API calls should use retry logic (`src/utils/retry.ts`). - MCP servers are auto-installed at runtime to `~/.claude/mcp/github-{type}-server
Model Context ProtocolCLAUDE.mdmodelcontextprotocol/registry/CLAUDE.md
humans (probably most things!), put the instructions in README.md instead._ Import @README.md ## Important: Publishing MCP servers The `data/seed.json` file is seed data for local development only. Do NOT create pull
grabCLAUDE.mdgrab/cursor-talk-to-figma-mcp/CLAUDE.md
Cursor AI IDE with Figma. Three components communicate in a pipeline: ``` Cursor AI ←(stdio)→ MCP Server ←(WebSocket)→ WebSocket Relay ←(WebSocket)→ Figma Plugin ``` ## Build & Development Commands ```bash bun install # Install dependencies
airweave-aiCLAUDE.mdairweave-ai/airweave/CLAUDE.md
TypeScript, Vite, ShadCN UI, TailwindCSS - **Workers**: Temporal for async sync orchestration, Redis for pub/sub - **MCP Server** (`mcp/`): Node.js, Streamable HTTP transport for AI assistant integration Data flow: Sources → Entity extraction
op7418CLAUDE.mdop7418/CodePilot/CLAUDE.md
只从当前 Worktree 启动，不得在其他目录启动 - 合并回主分支必须由用户主动发起，不得自动合并 - **端口隔离**：Worktree 启动 dev server 时使用非默认端口（如 `PORT=3001`），避免与主目录冲突 - **禁止跨目录编辑**：属于 Worktree 任务范围的文件，只在该 Worktree 内编辑，不得在主目录修改 - **合并前检查 untracked 文件**：合并回主分支前先 `git status
SentryCLAUDE.mdgetsentry/XcodeBuildMCP/CLAUDE.md
applicable skills with `warden --skill `. - ## Multi-process filesystem state - MobileBuildMCP explicitly supports multiple concurrent MCP server, daemon, CLI, test, and helper processes for the same or different workspaces. - Shared filesystem
czlonkowskiCLAUDE.mdczlonkowski/n8n-skills/CLAUDE.md
teach AI assistants how to build flawless n8n workflows using the n8n-mcp MCP server. **Repository**: https://github.com/czlonkowski/n8n-skills **Purpose**: 14 complementary skills that provide expert guidance on using
tradesdontlieCLAUDE.mdtradesdontlie/tradingview-mcp/CLAUDE.md
study by default (pass `max_labels` to override) ## Architecture ``` Claude Code ←→ MCP Server (stdio) ←→ CDP (localhost:9222) ←→ TradingView Desktop (Electron) ``` Pine graphics path: `study._graphics._primitivesCollection.dwglines.get('lines').get(false)._primitivesDataById
jacob-bdCLAUDE.mdjacob-bd/gemini-notebook-mcp-cli/CLAUDE.md
when working with code in this repository. ## Project Overview **Gemini Notebook (formerly Google NotebookLM) MCP Server & CLI** - Provides programmatic access to Gemini Notebook (notebook.google.com) via both a Model Context Protocol
FlorianBruniauxCLAUDE.mdFlorianBruniaux/claude-code-ultimate-guide/CLAUDE.md
security/ # security-hardening, sandbox-isolation, sandbox-native, production-safety, data-privacy ├── ecosystem/ # ai-ecosystem, mcp-servers-ecosystem, third-party-tools, remarkable-ai ├── roles/ # ai-roles, adoption-approaches, learning-with
ZSeven-WCLAUDE.mdZSeven-W/openpencil/CLAUDE.md
desktop` (live MCP on `127.0.0.1: /mcp`) - **CLI:** `cargo build -p op-cli` → binary `op` - **MCP server:** built into the desktop/web host (`--mcp `); crate `op-mcp` - **Iconify catalog (Rust assets):** from
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.
