An integration that allows LLMs to interact with Raindrop.io bookmarks using the Model Context Protocol (MCP).
https://github.com/hiromitsusasaki/raindrop-io-mcp-serverYour Raindrop.io collection sits there, packed with resources you've saved but rarely revisit. Meanwhile, you're asking Claude to help with research, project planning, and technical decisions—completely disconnected from the knowledge you've already curated.
This MCP server fixes that disconnect. Now Claude can directly search, create, and organize your Raindrop.io bookmarks without you ever leaving the conversation.
You've built a knowledge base in Raindrop.io over months or years. Articles about React patterns, API documentation, design inspiration, competitor analysis—it's all there. But when you're working with Claude on a project, you end up either:
It's a workflow killer, especially during deep research sessions or when onboarding to new projects.
With this MCP server running, Claude becomes your bookmark-aware research partner:
Research Mode: "Find all my bookmarks tagged 'microservices' and 'postgres' from the last 6 months, then analyze the common patterns"
Project Planning: "Create bookmarks for these 5 API docs we discussed, tag them 'project-phoenix', and add them to my work collection"
Knowledge Discovery: "Search my bookmarks for anything related to React testing, then suggest which articles I should review for our upcoming refactor"
The conversation flows naturally—no tab switching, no copy-pasting URLs, no breaking your thought process.
The fastest path is through Smithery:
npx -y @smithery/cli install @hiromitsusasaki/raindrop-io-mcp-server --client claude
That's it. Smithery handles the Claude Desktop configuration automatically.
For manual setup, you'll need your Raindrop.io API token (grab it from your Raindrop.io integrations page), then add this to your Claude Desktop config:
{
"mcpServers": {
"raindrop": {
"command": "node",
"args": ["path/to/build/index.js"],
"env": {
"RAINDROP_TOKEN": "your_token_here"
}
}
}
}
create-bookmark: Save links with proper tagging and organization as you discover them in conversation. No more "I'll bookmark that later" and forgetting.
search-bookmarks: Query your entire collection with natural language. Find resources by content, tags, or timeframe without remembering exact titles.
Both tools work exactly how you'd expect—Claude understands your intent and handles the API calls seamlessly.
Due Diligence: You're evaluating a new framework. Ask Claude to search your existing bookmarks for related technologies, then create new bookmarks for the documentation and tutorials you find together.
Team Onboarding: "Find all my bookmarks tagged 'company-standards' and create a summary document with the key resources new developers should review."
Competitive Research: Search bookmarks by competitor names, then have Claude analyze patterns in the features and strategies you've been tracking.
Learning Path Creation: "Look through my 'frontend' bookmarks and suggest a learning sequence for someone moving from jQuery to React."
The server gives Claude direct access to your curated knowledge, making every conversation more informed and productive.
Built with the MCP SDK, proper TypeScript types, and following MCP best practices. The codebase is clean and extensible if you want to add custom functionality.
Integration is standard MCP—no proprietary protocols or vendor lock-in. If you switch AI assistants later, your setup stays relevant.
Check out the repository to see the implementation or contribute improvements. At 38 stars and growing, it's already helping developers bridge the gap between their bookmark collections and AI workflows.
Your carefully curated bookmarks shouldn't be trapped in a separate app. Make them part of your AI-assisted development process.