Skip to main content
Skip to main content
Edit this page

How to build a LangChain/LangGraph AI agent using ClickHouse MCP Server

In this guide, you'll learn how to build a LangChain/LangGraph AI agent that can interact with ClickHouse's SQL playground using ClickHouse's MCP Server.

Example notebook

This example can be found as a notebook in the examples repository.

Prerequisites​

  • You'll need to have Python installed on your system.
  • You'll need to have pip installed on your system.
  • You'll need an Anthropic API key, or API key from another LLM provider

You can run the following steps either from your Python REPL or via script.

Install libraries​

Install the required libraries by running the following commands:

Setup credentials​

Next, you'll need to provide your Anthropic API key:

Using another LLM provider

If you don't have an Anthropic API key, and want to use another LLM provider, you can find the instructions for setting up your credentials in the Langchain Providers docs

Initialize MCP Server​

Now configure the ClickHouse MCP Server to point at the ClickHouse SQL playground:

Configure the stream handler​

When working with Langchain and ClickHouse MCP Server, query results are often returned as streaming data rather than a single response. For large datasets or complex analytical queries that may take time to process, it's important to configure a stream handler. Without proper handling, this streamed output can be difficult to work with in your application.

Configure the handler for the streamed output so that it's easier to consume:

Call the agent​

Finally, call your agent and ask it who's committed the most code to ClickHouse:

You should see a similar response as below: