# chatmcp/mcp-server-chatsum [Health: Active]

**Category:** 💬 Communication  
**Repository:** https://github.com/chatmcp/mcp-server-chatsum  
**GitHub Stars:** 1026  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/chatmcp-mcp-server-chatsum

## Description
Query and summarize your chat messages with AI prompts.

## Tools
Capabilities this server exposes over MCP:

- **query_chat_messages** — Query chat messages

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What chatmcp/mcp-server-chatsum does

The chatmcp/mcp-server-chatsum MCP server gives an MCP client access to chat messages saved by the repository’s companion chatbot. Its purpose is to retrieve messages using supplied query parameters and produce a summary based on the query prompt.

The repository expects the chatbot component to be configured and running first. That component saves conversations into a chat database, which this server then reads. The server is therefore intended for environments where the relevant message history is already being collected by the companion application.

## How it works

The server communicates with MCP clients through standard input and output. A client invokes `query_chat_messages` with query parameters, and the tool returns matching chat content together with a summary generated from the supplied prompt.

Because the server reads an existing database rather than collecting messages itself, the chatbot setup is a prerequisite. Start the chatbot so it can save messages before using the MCP server to query them.

## Setup and configuration

To configure chatmcp/mcp-server-chatsum, create a `.env` file in the repository root and set `CHAT_DB_PATH` to the path of the chat database. The documented example points to the database under the chatbot’s data directory.

Install dependencies with `pnpm install`, then build the project with `pnpm build`. During development, `pnpm watch` enables automatic rebuilding. The resulting server entry point is `build/index.js`.

Claude Desktop can launch the server by supplying the Node executable as `command`, the built JavaScript file as an argument, and `CHAT_DB_PATH` in the server’s `env` object. The README identifies the configuration file locations for both macOS and Windows.

## Tools and capabilities

The server exposes one documented MCP tool:

- `query_chat_messages`: accepts query parameters to find chat messages and summarizes the results according to a query prompt.

The repository also does not document a separate resource or prompt interface.

## Limitations and notes

chatmcp/mcp-server-chatsum depends on the companion chatbot and its populated chat database. It does not appear to be a general-purpose chat connector; the provided setup specifically points to the database produced by that chatbot. You must provide a valid database path through `CHAT_DB_PATH`.

The documented transport is stdio, so the server is started as a local process by the MCP client. For troubleshooting, the repository recommends its `pnpm inspector` script, which launches the MCP Inspector and provides a browser URL for debugging.

_Full upstream README: https://allmcps.com/mcp/chatmcp-mcp-server-chatsum/readme_