Hydrolix MCP Server

What is Hydrolix MCP Server?

Hydrolix MCP Server is an MCP server that provides tools to query and explore Hydrolix clusters. It allows AI assistants like Claude Desktop and Claude Code to execute SQL queries, list databases, list tables, and retrieve table metadata from a Hydrolix data platform.

How to use Hydrolix MCP Server?

Install with uvx (recommended) or pip; requires Python 3.13+. Configure the server in your MCP client by providing HYDROLIX_URL plus either a username/password pair (HYDROLIX_USER and HYDROLIX_PASSWORD) or a service account token (HYDROLIX_TOKEN). After restarting the client, test with a prompt like "Using your Hydrolix MCP tools, list the available databases."

Key features of Hydrolix MCP Server

• Execute SQL queries via run_select_query tool
• List databases and tables on a Hydrolix cluster
• Retrieve table schema with get_table_info
• Supports multiple authentication methods (token or username/password)
• Health check endpoint at /health (HTTP/SSE transport)

Use cases of Hydrolix MCP Server

• Ask an AI assistant to list databases and explore table schemas
• Run ad‑hoc SQL queries on Hydrolix without a separate query tool
• Let AI generate and execute time‑range‑optimized queries against Hydrolix

FAQ from Hydrolix MCP Server

What tools does the Hydrolix MCP Server provide?

It provides run_select_query for SQL queries, list_databases to list databases, list_tables to list tables in a database, and get_table_info to retrieve table metadata.

How do I authenticate to Hydrolix?

The server supports per‑request bearer tokens, per‑request query‑parameter tokens, and environment‑based credentials (HYDROLIX_USER/HYDROLIX_PASSWORD or HYDROLIX_TOKEN). Authentication methods are tried in precedence order.

What are the runtime requirements?

Python 3.13 or later. The recommended launcher is uvx (via Astral’s uv), which manages dependencies automatically.

Can I use a health check?

Yes. When using HTTP or SSE transport, a health endpoint at /health returns 200 OK with the Hydrolix query‑head ClickHouse version, or 503 if the connection fails.

Where does my data live?

Data stays on your Hydrolix cluster; the MCP server only sends queries and receives results through the Hydrolix API.