mcpserver.so LogoMCPFly
Submit
Twitter Bookmarks MCP logo

Twitter Bookmarks MCP

by chrislee973

AI/Development ToolstwitterbookmarksClaudeSQLite

This MCP server provides Claude and other AI assistants access to your Twitter bookmarks in the form of a SQLite database. It allows you to search, analyze, chat with, and generate visualizations of your bookmarks directly in clients like Claude Desktop.

View on GitHub

Last updated: N/A

Twitter Bookmarks MCP

⚠️ This project is still very much a work in progress. Expect rough edges and significant changes. Feel free to open issues or reach out on twitter @626ripes if you encounter any problems.

This is a MCP server for providing Claude and other AI assitants access to your Twitter bookmarks in the form of a SQLite database so you can search, analyze, chat with, and generate visualizations of them directly in clients like Claude Desktop.

It's a companion for my Twitter Boomark Search app.

Available Tools

This MCP server is a fairly simple and thin interface layer over a read-only sqlite datbase. It has the below simple tools

  • get_schema(): Returns the database schema. This is called at the start of each conversation to understand the database structure.
  • run_query(sql): Executes read-only SQL queries.
  • search_text(query): Performs full-text search on the database's fts index. Returns complete tweet information including text, URLs, user details, and embedded links.

Setup

Prerequisites

  • Make sure you have the sqlite database file of your bookmarks. To do so, first extract your bookmarks in json format by following the steps here and upload them in that same page. Then go to the claude-plugin page to convert them to a sqlite database for you to download to your computer. Take note of the filepath of this saved sqlite file.
  • uv package manager

Installation

  1. Clone this repository:

    git clone https://github.com/chrislee973/twitter-bookmarks-mcp
cd twitter-bookmarks-mcp

  2. Install dependencies:

    uv venv && uv pip install -r pyproject.toml

Connecting to Claude Desktop

  1. Run the command:
uv run mcp install -e . server.py -v DB_PATH=path/to/db

Replace path/to/db with the full path to wherever your database file is saved.

  1. Restart Claude Desktop

  2. Try asking Claude questions like:

    • "Return the abstracts of my 5 most recent bookmarked arxiv papers"
    • "Look through my twitter bookmarks for bookmarks that contain a link to the blog lesswrong, and summarize the content of each of those blog posts"
    • "Create a chart of my bookmark frequency over time"