Skip to Content
User Guide🧩 Available MCP ServersClickHouse

ClickHouse

Connect to ClickHouse database to execute high-performance analytical queries.

Key Features

  • Execute high-performance analytical queries
  • Columnar database optimization
  • Large-scale data processing
  • Real-time data analysis
  • Aggregate and window function support
  • Read-only mode support

Basic Information

Authentication Method

Authenticate using username/password.

Input Parameters

  • Host (required): ClickHouse server host
  • Port (required): ClickHouse server port (default: 8123)
  • Username (required): ClickHouse username
  • Password (required): ClickHouse password
  • Database (optional): Database name to connect to (default: default)
  • Connection Timeout (optional): Connection timeout (default: 30000ms, max: 60000ms)
  • Use SSL (optional): Whether to use SSL connection
  • Read Only (optional): Whether to enable read-only mode

Supported MCP Tools

execute_sql

Execute SQL queries on ClickHouse database and return results.

In read-only mode, only the following SQL commands are allowed:

  • SELECT - Query data
  • WITH - Use Common Table Expression (CTE)
  • EXPLAIN - Query execution plan
  • ANALYZE - Query analysis
  • SHOW - Metadata queries

Result Limit: Returns a maximum of 500 rows for performance reasons.

For example,

  • When you input “Query the user table data” as a prompt,
  • The LLM executes the SELECT * FROM users query through MCP.
  • Then, it responds in the chat based on the tool call results.

ClickHouse Connection Setup

  1. Prepare a user account that can access ClickHouse server.
  2. Grant necessary database permissions.
  3. Configure network access permissions.
  4. Select ClickHouse integration in QueryPie AI.
  5. Enter connection information to complete the setup.

Precautions

  • Keep your ClickHouse user account password secure.
  • It’s recommended to grant only the minimum necessary database permissions.
  • It’s recommended to use SSL connections in production environments.
  • Large query results are limited to 500 rows, so use appropriate LIMIT clauses.
  • You can prevent data changes by enabling read-only mode.
Last updated on
Oracle DatabaseSnowflake