R

Raindrop.io

Read your Raindrop.io bookmarks and collections.

Works with: Claude DesktopCursor
Quick install
npx -y raindrop-mcp

How to install the Raindrop.io MCP server

Add this to your Claude Desktop MCP configuration:

{
  "mcpServers": {
    "raindrop": {
      "command": "npx",
      "args": [
        "-y",
        "raindrop-mcp"
      ]
    }
  }
}

Add this to your Cursor MCP configuration:

{
  "mcpServers": {
    "raindrop": {
      "command": "npx",
      "args": [
        "-y",
        "raindrop-mcp"
      ]
    }
  }
}

The Raindrop.io MCP server gives Claude access to your bookmark library. Search across collections, tags, and highlights. For long-time Raindrop users, this turns years of saved articles into something Claude can reason over.

Raindrop is one of the most popular dedicated bookmark managers, especially for users who want a tag-and-collection structure rather than a flat folder list. The MCP server is a natural fit for that audience.

Why use it

Bookmarks rot when they sit in a list. The whole point of saving them is to surface them at the right moment, but that almost never happens with a manual UI. The MCP server is what closes that loop. Ask a question, Claude pulls relevant bookmarks, you get the answer with sources you’ve already trusted enough to save.

For research-heavy roles where you’ve been collecting links for years, this is the highest-leverage use of an existing Raindrop library.

What it actually does

Search bookmarks by query, tag, collection, or type. Fetch a specific bookmark with its content excerpt, tags, and highlights. List collections. Some servers also expose write endpoints for creating bookmarks from Claude.

Practical patterns:

  • “Find every Raindrop bookmark I tagged ‘design-systems’ last year.”
  • “Pull my collection ‘AI papers’ and summarize the top five by my own highlights.”
  • “What did I save about cold email outreach? Use Raindrop, not the live web.”

Gotchas

API rate limits are reasonable but heavy queries (especially across all collections) can take a while. The server handles pagination automatically but Claude may pause mid-response if results are large.

Highlights are Pro-only. If you’re on free Raindrop, the highlights field will be empty in API responses. Plain bookmark metadata still works.

For a more comprehensive memory setup, run Raindrop alongside ContextBolt. Raindrop captures what you save manually from your browser. ContextBolt captures every X, Reddit, and LinkedIn post you bookmark, automatically. Both feed Claude. Together they cover where most knowledge actually lives for solo founders and creators.

If you also use Readwise, you have a third source: long-form highlights from books and Reader. The combination makes Claude very good at “what have I already learned about X?” questions.

Also in Memory & Knowledge

Combine Raindrop.io with ContextBolt

Raindrop.io gives Claude one kind of memory. ContextBolt adds another: every tweet, post, and article you save across X, Reddit, and LinkedIn becomes searchable by meaning. Run both as MCP servers and Claude can pull from both layers in one prompt.

See ContextBolt →

Raindrop.io MCP server: FAQs

Is the Raindrop server official?

No. Raindrop.io hasn't shipped a first-party MCP server. The community implementations wrap the public API and work for read use cases. Authentication is a personal access token from the Raindrop settings page.

Does it support all my collections?

Yes. Standard collections, nested collections, and the 'Unsorted' inbox are all queryable. Most servers also expose tags, types (article, video, image), and the highlights you've made on bookmarked pages.

Can it create bookmarks too?

Some servers support write endpoints, but most are read-only. If you want Claude to bookmark URLs from a conversation into Raindrop, check the specific server you install.

How is this different from ContextBolt?

Raindrop is a manual save tool. You hit the bookmark button, it goes in. ContextBolt auto-saves what you bookmark on X, Reddit, and LinkedIn without leaving the platform. They overlap on intent but not on input. Many people use both, with Raindrop for browser-found articles and ContextBolt for social.

What about the Raindrop highlights?

The Pro version of Raindrop lets you highlight passages in saved articles. The MCP server can return these. So you can ask Claude things like 'find every highlight I made in articles about pricing' and it will pull the underlying snippets.