What is Twitter Bookmark Mcp

twitter-bookmark-mcp is a server designed to provide AI assistants like Claude access to your Twitter bookmarks through a SQLite database, allowing for search, analysis, and visualization of bookmarks.

Use cases

Use cases for twitter-bookmark-mcp include searching for specific tweets among bookmarks, generating visualizations of saved content, and integrating with AI assistants for enhanced interaction with Twitter data.

How to use

To use twitter-bookmark-mcp, you need to first extract your Twitter bookmarks into a JSON format, convert them into a SQLite database, and then run the MCP server with the path to your database file. You can interact with the server using commands like get_schema(), run_query(sql), and search_text(query).

Key features

Key features include a simple interface over a read-only SQLite database, the ability to retrieve database schema, execute SQL queries, and perform full-text searches on bookmarks.

Where to use

twitter-bookmark-mcp can be used in areas such as data analysis, research, and personal organization, particularly for users who want to manage and analyze their Twitter bookmarks effectively.

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.

2. Restart Claude Desktop

3. 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”