Docs MCP Server
Grounded Docs MCP Server
A local MCP server that indexes docs from websites, GitHub, npm, PyPI and files and gives the agent version-aware search
- 1,875 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs a local server and scrapes the named external sites
- Indexes local folders and files, and with a provider key sends text out for embeddings
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/grounded-docs-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs a local server and scrapes the named external sites
- Indexes local folders and files, and with a provider key sends text out for embeddings
This is third-party code. Review the repository files before installing.
What it does
The server collects documentation from external sources and local files into its own index on your machine, then serves search over it to the agent through MCP. Sources include websites, GitHub repositories, npm and PyPI packages, local folders and zip archives, and it reads not only HTML and Markdown but also PDF, office formats, EPUB, Jupyter notebooks and source code in many languages. A query can target a specific library version so answers rest on real docs rather than model memory. A web interface on port 6280 adds and manages sources, and a CLI indexes and searches from scripts. An optional embedding model enables semantic search.
Who it is for. For developers who want their AI assistant to answer from the current docs of their libraries and versions.
Good fit when
- The agent returns outdated or invented APIs and you need answers from real docs
- You need to index a specific library version's documentation and search it
- You want a local documentation index without sending code to an external service
Not a fit when
- A one-off browser search is enough without running a server
- You cannot keep a local index or run Node.js or Docker
Example request
Index the React docs from react.dev and find the correct way to clean up an effect in useEffectMCP
- Transport
- stdio, http, sse
- Authentication
- not required
Environment variables
OPENAI_API_KEYsecret- OpenAI key for semantic search via embeddings. Optional; other providers and local Ollama models are supported.
Limitations
The index and server run locally: you start and maintain them yourself. Running via npx needs Node.js 22 or newer, with a Docker image as an alternative. Without an embedding model search stays lexical; semantic search needs an OpenAI or other provider key, or a local model via Ollama. Sending text to OpenAI for embeddings goes to an external service.
How to disable. Stop the server process or the Docker container and remove the docs-mcp-server entry from your client's MCP config. The local index database can be deleted manually.
FAQ
Is an OpenAI key required?
No. Lexical search works without an embedding model. For semantic search you can connect OpenAI, another provider or a local model via Ollama.
Does code or data leave the machine?
The server and index run on your machine. Only scraping the named sources and, if enabled, sending text to the chosen embedding provider goes out.
Discussion
No questions or reviews yet.
Related
MCP serversRef's MCP server: documentation search for APIs and libraries that hands the agent only the relevant parts of pages
A remote MCP server that gives the agent docs and code of any public GitHub repository with nothing to install
MCP server that fetches the exact documentation section instead of the whole file: a heading index and byte-precise reads
A Go crawler turns documentation sites into Markdown and gives agents offline search_docs and page reading over MCP.
An MCP server that turns a large codebase into a semantic feature graph for precise navigation and meaning-based search
mcpdoc
MCP LLMS-TXT Documentation Server
An MCP server from LangChain that serves docs from the llms.txt files you choose, with auditable tool calls
README in short
The README describes the server as an open alternative to Context7, Nia and Ref.Tools that fights outdated and hallucinated AI answers by providing an index of current docs. It pulls documents from websites, GitHub, npm, PyPI and local files and supports many formats, from PDF and office documents to source code. For agents and scripts the CLI is simplest: the scrape, search and fetch-url commands. For a persistent endpoint it runs a server with a web interface on port 6280, connected by Claude, Cline, Copilot, Gemini CLI and other MCP clients over stdio, http or sse. An embedding model is optional and improves search quality. MIT licensed.