ref-tools-mcp

Overview

Ref MCP is a Model Context Protocol server that enables AI coding agents to efficiently search and retrieve documentation for APIs, libraries, and services. The server uses innovative techniques to minimize token usage by filtering search results and extracting only the most relevant sections of documentation pages. This approach helps reduce context rot and control API costs by avoiding the inclusion of unnecessary tokens.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• ref_search_documentationquery: string

  A powerful search tool to check technical documentation.

  Try:Search for documentation on the Figma API post comment endpointTry:Find information about n8n merge node best practicesTry:Search for Python async/await documentation
• ref_read_urlurl: string

  Fetches content from a URL and converts it to markdown.

  Try:Read the content from https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.merge/Try:Fetch documentation from https://api.example.com/docsTry:Convert https://github.com/user/repo/blob/main/README.md to markdown

Comparable tools

Installation

"Ref": {
  "type": "http",
  "url": "https://api.ref.tools/mcp?apiKey=YOUR_API_KEY"
}

"Ref": {
  "command": "npx",
  "args": ["ref-tools-mcp@latest"],
  "env": {
    "REF_API_KEY": <sign up to get an api key>
  }
}

{
  "mcpServers": {
    "ref-tools": {
      "command": "npx",
      "args": ["ref-tools-mcp@latest"]
    }
  }
}

FAQ

How does Ref MCP reduce token usage?

Ref MCP filters search results to avoid duplicates and extracts only the most relevant 5k tokens from documentation pages, instead of fetching entire pages which could contain 20k+ tokens of mostly irrelevant content.

What documentation sources does Ref support?

Ref supports searching public documentation on the web, GitHub repositories, and private resources like repositories and PDFs. It can fetch content from any URL provided by its search results.

Compare ref-tools-mcp with