Trino MCP Server

Trino MCP Server is an MCP server that connects AI assistants to the Trino distributed SQL query engine. Built using the official Python Model Context Protocol library and trino.dbapi client, it enables data engineers, analysts, and developers to explore distributed data architectures directly through language models. The server exposes Trino tables as native MCP resources, lets users inspect table contents, and provides an execution tool for running arbitrary SQL queries across connected catalogs and schemas. Instead of manually switching to SQL workbenches or command-line interfaces, teams can connect tools like Claude Desktop or Cursor to federate analytical queries across data warehouses, object storage, and relational databases. It handles authentication and connection routing through standard environment variables, making it straightforward to deploy alongside existing Trino infrastructure for ad-hoc analytical workflows and automated data discovery.

Category: Data & Analytics

Tags: data-warehousing, distributed-query, SQL, trino

Visit Trino MCP Server

How to install and configure Trino MCP Server

1. Ensure Python 3.9+ is installed along with the mcp and trino Python packages. 2. Clone the repository locally using git: git clone https://github.com/Dataring-engineering/mcp-server-trino. 3. Add the server definition to your MCP client configuration (such as Claude Desktop or Cursor) using the following JSON snippet: json { "mcpServers": { "trino": { "command": "uv", "args": [ "--directory", "<path_to_mcp_server_trino>", "run", "mcp_server_trino" ], "env": { "TRINO_HOST": "<host>", "TRINO_PORT": "<port>", "TRINO_USER": "<user>", "TRINO_PASSWORD": "<password>", "TRINO_CATALOG": "<catalog>", "TRINO_SCHEMA": "<schema>" } } } } 4. Replace <path_to_mcp_server_trino> with your local directory path and specify your connection credentials before restarting your client.

What you can do with Trino MCP Server

• Discovering schemas and listing tables across federated Trino catalogs through automated natural language prompts. - Inspecting table schemas and reading sample records to understand distributed data structures before running larger pipelines. - Executing analytical SQL queries across multiple connected data sources to aggregate metrics directly within an AI chat interface. - Troubleshooting data pipeline failures by querying underlying lakehouse or warehouse tables using ad-hoc diagnostic SQL commands.

Key facts

• Open Source
• https://github.com/Dataring-engineering/mcp-server-trino
• Data & Analytics, Databases & Data Stores
• data-warehousing, distributed-query, SQL, trino

Part of MCP Servers

How do I install Trino MCP Server?

To run the server, clone the repository from GitHub and ensure you have Python 3.9 or higher installed alongside the trino and mcp packages. Register the server in your MCP client configuration file by pointing the uv command to the cloned directory and setting environment variables for your host, port, user, catalog, and schema.

What can Trino MCP Server do?

The server exposes Trino tables as discoverable MCP resources, allows reading raw table contents via the protocol, and offers a tool to execute arbitrary SQL queries against your Trino clusters, catalogs, and schemas.

Which MCP clients work with Trino MCP Server?

Any client supporting the Model Context Protocol over standard input and output can use this server, including Claude Desktop and AI code editors like Cursor configured with local command execution.

What configuration variables are required?

You must configure TRINO_USER, TRINO_CATALOG, and TRINO_SCHEMA via environment variables. Optional settings include TRINO_HOST (defaults to localhost), TRINO_PORT (defaults to 8080), and TRINO_PASSWORD depending on your cluster authentication setup.

Is Trino MCP Server open source?

Yes, Trino MCP Server is an open-source project hosted on GitHub under the Dataring-engineering organization.