ClickHouse MCP Server

A Model Context Protocal (MCP) server implement for ClickHouse.

This server provides AI assistants with a secure and structured way to explore and analyze databases. It enables them to list tables, read data, and execute SQL queries through a controlled interface, ensuring responsible database access.

Configuration

Set the following environment variables:

CLICKHOUSE_HOST=localhost    
CLICKHOUSE_PORT=8123         
CLICKHOUSE_USER=default
CLICKHOUSE_PASSWORD=CHANGE_TO_YOUR_PASSWORD
CLICKHOUSE_DATABASE=default

Or via command-line args:

--host the database host
--port the database port
--user the database username
--password the database password
--database the database name

Usage

Fake some data in clickhouse

Check the SQL in dev_contribute.md for details.

Post your question to AI assistant in Cline

What is the sales volume in each region? Which product is the best - selling?

CLINE

Configure the MCP server in VSCode, Cline extension, or other MCP client.: Example:

{
  "mcpServers": {
    "clickhouse": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/clickhouse_mcp_server",
        "run",
        "-m",
        "clickhouse_mcp_server.server"
      ],
      "env": {
        "CLICKHOUSE_HOST": "localhost",
        "CLICKHOUSE_PORT": "8123",
        "CLICKHOUSE_USER": "default",
        "CLICKHOUSE_PASSWORD": "CHANGE_TO_YOUR_PASSWORD",
        "CLICKHOUSE_DATABASE": "default"
      }
    }
    
  }
}

License

APACHE - See LICENSE file for details.

Contribute

See dev_contribute.md for details.

Prerequisites

Python with uv package manager
ClickHouse installation
MCP server dependencies

Acknowledgement

This library's implementation was inspired by the following three repositories and incorporates their code, respect for the open-source spirit!

Thanks!