Bookmark Geni

Bookmark Geni MCP Server is a stand‑alone MCP (Model Context Protocol) server designed to index local browser bookmarks, fetch and extract page content, and provide fast, natural‑language search over your bookmarks. It reads multiple browser bookmark formats, enriches them with metadata, and stores vector embeddings to enable semantic search with optional filtering.

How to use

You run the server locally and connect to it with an MCP client or integration. Index your bookmarks from the supported browsers, let the server fetch HTML content and extract metadata, and then perform natural‑language queries over your bookmarks. You can filter results by metadata and control how many results are returned.

How to install

Prerequisites: you need Python and the npm/yarn tooling if you plan to install additional dependencies locally. Ensure you have pip available to install Python requirements.

Step 1: Install dependencies.

Additional sections

Configuration: you can tune which browsers are enabled, where embeddings are stored, and how many URLs to process. The server reads settings from a configuration file named config.yaml in the MCP server root. You can enable or disable browsers, set the ChromaDB path, and define the metadata storage location.

Security: avoid exposing the server to untrusted networks. Run the MCP server behind a firewall or VPN and manage access through your MCP client configuration.

Notes: the server supports batch processing with concurrency and caching for performance. It can export and import embeddings to and from pickle files for easy transfer.

Example configurations

{
  "mcpServers": {
    "bookmarkGeni": {
      "type": "stdio",
      "command": "bash",
      "args": ["/path/to/bookmark_geni_mcp/scripts/start_mcp_server.sh"],
      "env": {
        "PYTHON_PATH": "/path/to/your/python3"
      }
    }
  }
}

Starting the server locally

chmod +x scripts/start_mcp_server.sh
./scripts/start_mcp_server.sh

Available tools

generate_bookmarks_metadata

Scans selected browsers for bookmarks, fetches HTML content, extracts metadata, creates embeddings, and stores them in the vector store.

query_bookmarks

Performs semantic search over stored bookmarks with optional metadata filtering and result limiting.

list_browsers

Lists installed browsers and their detected bookmark file paths.

get_stats

Returns database statistics such as total bookmarks and collection info.

export_embeddings

Exports all embedding data to a pickle file for backup or transfer.

import_embeddings

Imports embedding data from a pickle file.