LinksBubble ships with a built-in Model Context Protocol (MCP) server. This allows AI coding assistants like Claude Desktop and Cursor to securely read your saved links, notes, and tags locally, providing deep context for your conversations.
Capabilities
- Semantic Search: Ask Claude to "find my notes on React performance" and the MCP server will query your local library.
- Direct Read: The assistant can instantly fetch the full parsed text of any URL you have captured.
- Tag Exploration: AI can traverse your tag hierarchy to build context on specific projects.
How to Connect (Automatic)
For officially supported tools (Cursor IDE, Claude Desktop, and Claude Code), setup is completely automatic:
- Open the LinksBubble app on your desktop.
- Click the Options menu (...) in the top right corner.
- Click the Connect button under the MCP section.
- Restart Cursor or Claude Desktop. That's it!
- Usage Example: Open Claude or Cursor and type: "Search my LinksBubble for 'Architecture' and summarize the results." You will see it automatically use the LinksBubble MCP tool to read your saved items!
How to Connect (Manual / Other Tools)
For unofficial VS Code extensions (like Cline or Roo Code) or other tools, use the manual setup:
- Click the Copy icon next to the Connect button in LinksBubble to copy your unique MCP server path.
- Open your AI extension's MCP configuration settings.
- Add the following configuration, pasting your copied path into the command field:
"mcpServers": {
"linksbubble": {
"command": "PASTE_YOUR_COPIED_PATH_HERE",
"args": []
}
}