claudekeep
ClaudeKeep lets you save & share chats from Claude using an MCP inside Claude Desktop
claude mcp add --transport stdio sdairs-claudekeep npx -y claudekeep-mcp \ --env CLAUDEKEEP_TOKEN="<YOUR_TOKEN>"
How to use
ClaudeKeep is an experiment using the Model Context Protocol (MCP) to save and share your AI conversations from Claude Desktop. The MCP server implementation enables Claude to save your chats via a dedicated tool, and a companion web app lets you view private chats and public chats. To use it, you first authenticate with Claude Keep to obtain a token. Then configure Claude Desktop to point at the MCP server so it can store chats you choose to save. The workflow is designed so that every message you write can be stored locally in Claude Desktop, and when you ask Claude to save, the chat is stored with privacy controls (private by default, with an option to publish). The server is intended as an experimental tool for managing chat context rather than a production-grade solution, so treat sensitive information with care.
How to install
Prerequisites:
- Node.js and npm installed on your machine (or access to a system where Node.js is available).
- Basic familiarity with MCP configuration in Claude Desktop.
Step-by-step installation:
- Ensure Node.js is installed. You can verify with:
node -v
npm -v
- Install and run the MCP server via npx (no local installation required):
npx -y claudekeep-mcp
-
Obtain and provide your CLAUDEKEEP_TOKEN when configuring Claude Desktop. The token is obtained from claudekeep.com after logging in with GitHub as described in the project notes.
-
Configure Claude Desktop with the MCP server entry:
- On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- On Windows: %APPDATA%/Claude/claude_desktop_config.json
Example configuration snippet to paste into your Claude Desktop config:
{
"mcpServers": {
"claudekeep-mcp": {
"command": "npx",
"args": [
"-y",
"claudekeep-mcp"
],
"env": {
"CLAUDEKEEP_TOKEN": "<YOUR_TOKEN>"
}
}
}
}
- Restart Claude Desktop completely (kill and relaunch). Then select the MCP in Claude Desktop to begin saving chats using the store/save tools as described in the project notes.
Additional notes
Tips and considerations:
- Your token is sensitive. If exposed, you can refresh it from the claudeKeep portal; reconfigure Claude Desktop with the new token.
- Public chats may be visible to others depending on how you save and publish. By default, chats are stored as private unless you explicitly request public visibility.
- The server is experimental and not guaranteed to be secure or perfect. Avoid storing highly sensitive data.
- If Claude Desktop has trouble reading PATH, ensure the MCP command is resolvable via your shell environment or specify the full path where applicable.
- Public chats are associated with your GitHub account on the server, so while individual chats are anonymous to other users, server-side associations exist.
Related MCP Servers
python-notebook
Lightweight Python Notebook MCP - Enable AI assistants to create, edit, and view Jupyter notebooks via Model Context Protocol
docmole
Dig through any documentation with AI - MCP server for Claude, Cursor, and other AI assistants
mongo
MCP server that provide tools to LLMs such as claude in cursor to interact with MongoDB
obsidian
MCP server for Obsidian vault management - enables Claude and other AI assistants to read, write, search, and organize your notes
gmail
A robust Model Context Protocol server for Gmail integration with intelligent authentication and comprehensive email operations
harvest
MCP server for Harvest time tracking API - use with Claude Desktop and other AI assistants