ref-tools MCP Server

Use This MCP server To

Look up how a library's feature actually works before writing code Check the latest documentation for a tool I've never used Find the exact example from an official guide Search my own private repositories or PDFs for the right answer Get accurate answers instead of outdated guesses about how software works Pull the right page from a long manual without reading the whole thing

README

License

Ref MCP

A ModelContextProtocol server that gives your AI coding tool or agent access to documentation for APIs, services, libraries etc. It's your one-stop-shop to keep your agent up-to-date on documentation in a fast and token-efficient way.

For more see info ref.tools

Agentic search for exactly the right context

Ref's tools are designed to match how models search while using as little context as possible to reduce context rot. The goal is to find exactly the context your coding agent needs to be successful while using minimum tokens.

Depending on the complexity of the prompt, LLM coding agents like Claude Code will typically do one or more searches and then choose a few resources to read in more depth.

For a simple query about Figma's Comment REST API it will make a couple calls to get exactly what it needs:

SEARCH 'Figma API post comment endpoint documentation' (54 tokens)
READ https://www.figma.com/developers/api#post-comments-endpoint (385 tokens)

For more complex situations, the LLM will try to refine it's prompt as it reads results. For example:

SEARCH 'n8n merge node vs Code node multiple inputs best practices' (126)
READ https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.merge/#merge (4961)
READ https://docs.n8n.io/flow-logic/merging/#merge-data-from-multiple-node-executions (138)
SEARCH 'n8n Code node multiple inputs best practices when to use' (107)
READ https://docs.n8n.io/code/code-node/#usage (80)
SEARCH 'n8n Code node access multiple inputs from different nodes' (370)
SEARCH 'n8n Code node $input access multiple node inputs' (372)
READ https://docs.n8n.io/code/builtin/output-other-nodes/#output-of-other-nodes (2310)

Ref takes advantage of MCP sessions to track search trajectory and minimize context usage. There's a lot more ideas cooking but here's what we've implemented so far.

1. Filtering search results

ref-tools-mcp FAQ

Can I use this to look up documentation for a library I'm unfamiliar with?

Yes — it searches public documentation sites and returns the exact section your assistant needs, so it can write code that matches how the tool actually works today.

Which apps does this work in?

It's built for Cursor and VS Code, and it also works with any assistant that supports the Model Context Protocol, including Claude Desktop.

Do I need an account or API key?

Yes — you'll need to sign up at ref.tools for a free API key before you can use it.

How hard is it to set up?

It's a short copy-and-paste step — you add a small snippet with your API key to your app's settings, and no separate installation is required if you use the hosted version.

Can I use this to search documents I don't want made public, like a private code repository?

Yes — beyond public documentation, it can be pointed at your own private repositories or PDFs.

Does this work with ChatGPT, Claude, or Gemini?

Yes — it works as a general documentation search tool for assistants like Claude, ChatGPT, and Gemini, since it follows the shared Model Context Protocol standard.

Will this slow down my coding assistant or use up a lot of its memory?

No — it's built to return only the most relevant few pages of information instead of an entire manual, which keeps your assistant's responses fast and focused.