docfork
Docfork - Up-to-date Docs for AI Agents.
claude mcp add --transport stdio docfork-docfork npx -y docfork --api-key YOUR_API_KEY
How to use
Docfork MCP provides a cabinet-backed context for AI agents. It locks the agent’s documentation context to your stack (Next.js 16 + Drizzle ORM + Better Auth) and serves only relevant docs from your libraries, enabling precise, fast, and non-hallucinatory results. With the Docfork MCP server, you can query for documentation and code examples that are scoped to your verified dependencies, and you can share Cabinets and API keys across your organization to ensure consistent context across teammates. The MCP exposes tooling that lets you fetch and search docs tailored to your stack, so you can retrieve exact code snippets and references without sifting through unrelated material. Typical workflows involve initializing the MCP server with your API key, then issuing prompts that leverage the search_docs and fetch_doc capabilities to locate and extract the most relevant docs for a given library or topic.
The available capabilities include:
- search_docs: quickly locate documentation results within your cabinet-scoped libraries based on keywords or library names.
- fetch_doc: retrieve full documentation content and relevant code samples for a selected result.
- Cabinet management: define and share Cabinets to ensure all agents reference the same stack and dependencies. To get started, obtain a Free API Key from Docfork, run the local MCP client to connect (or use a remote config), and begin issuing prompts that leverage the Docfork tools to fetch precise docs and code examples relevant to your project’s stack.
How to install
Prerequisites:\n- Node.js 16+ (LTS) and npm/yarn installed.\n- An active Docfork API key (free tier available).\n\nStep-by-step installation:\n1) Install Node.js if you don’t have it:\n - macOS: brew install node\n - Windows: download from nodejs.org and run the installer\n - Linux: use your distro’s package manager (e.g., apt/yum) or nvm to install Node.js.\n\n2) Install the Docfork MCP client (global or per-project):\n - Global (recommended for quick access):\n npm install -g docfork\n - Local to a project:\n npm init -y && npm install docfork --save-dev\n\n3) Run the MCP server client with your API key:\n - If installed globally:\n npx -y docfork --api-key YOUR_API_KEY\n - If installed locally:\n npx -y docfork --api-key YOUR_API_KEY\n\n4) (Optional) Connect via Cursor/OpenCode/Cline/OpenOpenCode as per your workflow by following their MCP config examples and substituting YOUR_API_KEY.\n\nPractical note: The command above launches the Docfork MCP client which exposes a server-side endpoint that your prompts can interact with. Ensure your API key is kept secure and do not commit it to public repositories.
Additional notes
Tips and common issues:\n- If you encounter authentication errors, verify that YOUR_API_KEY is valid and has the necessary scopes for cabinet access.\n- For remote connections, you may need to provide an API key via header-based authentication; ensure the API key is sent correctly by your integration layer.\n- Use Cabinets to lock your agent to a specific dependency graph; changing dependencies without updating the cabinet may result in missing docs or mismatched references.\n- If docs seem incomplete, ensure your cabinet libraries are up to date and that the Docfork API key has access to the required libraries.\n- OAuth endpoint support is available for remote connections; switch to the oauth endpoint if your workflow requires it.
Related MCP Servers
context7
Context7 MCP Server -- Up-to-date code documentation for LLMs and AI code editors
web-eval-agent
An MCP server that autonomously evaluates web applications.
unitree-go2
The Unitree Go2 MCP Server is a server built on the MCP that enables users to control the Unitree Go2 robot using natural language commands interpreted by a LLM.
mcp-chain-of-draft
Chain of Draft Server is a powerful AI-driven tool that helps developers make better decisions through systematic, iterative refinement of thoughts and designs. It integrates seamlessly with popular AI agents and provides a structured approach to reasoning, API design, architecture decisions, code reviews, and implementation planning.
cco
Real-time audit and approval system for Claude Code tool calls.
mcpverified
The definitive resource for Model Control Protocol (MCP) - Opinionated. Curated. Verified.