Ref

A Model Context Protocol (MCP) server that provides token-efficient search over public and private documentation. Ref MCP is designed to stop hallucinations by giving your AI coding tool or agent access to accurate, up-to-date documentation for APIs, services, libraries, and frameworks in a fast and token-efficient way.

Features

• Agentic Documentation Search: Intelligent search designed to match how models search while using minimal context to reduce context rot
• Token Efficiency: Filters repeated search results in sessions and fetches only the most relevant 5k tokens from documentation pages
• Public & Private Docs: Search across web documentation, GitHub repositories, and private resources like repos and PDFs
• Smart Context Management: Uses MCP sessions to track search trajectory and minimize context usage
• URL Content Fetching: Convert web pages to markdown with intelligent section filtering based on search history

What you’ll need to connect

Required:

• Ref API Key: Sign up at ref.tools to get your API key