MCP

Connect Trino to AI Assistants

Name: Connect Trino to AI Assistants
Availability: OnlineOnly
Author: Tommy Nguyen

Trino MCP Server: A Go-based server enabling AI assistants to interact with Trino, a distributed SQL engine, via standardized MCP tools for data querying and

Connect

Works with trinopostgresqlmysqls3hive

Tommy Nguyen

Maintainer?

Spark score

out of 100

Updated 4 months ago

Version 1.0.0

Models

claude

Add to Favorites

Why it matters

Enable AI assistants to query and interact with your Trino data warehouse. This MCP server provides a standardized interface for accessing distributed SQL data.

Outcomes

What it gets done

Execute SQL queries against Trino

Discover catalogs, schemas, and tables

Retrieve table schema definitions

Obtain query execution plans

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/vb-trino | bash

Capabilities

Tools your agent gets

execute_query

Execute SQL queries in Trino

list_catalogs

List available catalogs in Trino

list_schemas

List schemas in catalogs

list_tables

List tables in schemas

get_table_schema

Get table schema definition

explain_query

Get query execution plan

Overview

Trino MCP Server

What it does

This is a high-performance Model Context Protocol (MCP) server for Trino, implemented in Go. It enables AI assistants to interact with the Trino distributed SQL engine through standardized MCP tools. The server supports various features including query execution, catalog/schema/table discovery, and OAuth 2.1 authentication.

How it connects

Use this server to allow AI assistants to query Trino SQL, discover data structures (catalogs, schemas, tables), and retrieve query execution plans. It is designed for integration with AI workflows that require access to Trino data. Do not use if interaction with Trino via MCP is not needed.

Source README

A high-performance Model Context Protocol (MCP) server for Trino, implemented in Go, that enables AI assistants to easily interact with the distributed SQL engine Trino through standardized MCP tools.

Installation

Homebrew

brew install tuannvm/mcp/mcp-trino

One-command Installation

curl -fsSL https://raw.githubusercontent.com/tuannvm/mcp-trino/main/install.sh | bash

Local Development

export TRINO_HOST=localhost TRINO_USER=trino
mcp-trino

Available Tools

Tool	Description
`execute_query`	Execute SQL queries in Trino
`list_catalogs`	List available catalogs in Trino
`list_schemas`	List schemas in catalogs
`list_tables`	List tables in schemas
`get_table_schema`	Get table schema definition
`explain_query`	Get query execution plan

Features

MCP server implementation in Go
Execute Trino SQL queries through MCP tools
Catalog, schema, and table discovery
Docker container support
STDIO and HTTP transport support
OAuth 2.1 authentication with 4 providers (HMAC, Okta, Google, Azure AD)
Native and proxy OAuth modes
StreamableHTTP support with JWT authentication
Backward compatibility with SSE endpoints
Compatibility with Cursor, Claude Desktop, Windsurf, ChatWise

Environment Variables

Required

TRINO_HOST - Trino server host
TRINO_USER - Trino username

Optional

TRINO_SCHEME - Connection scheme for Trino
MCP_TRANSPORT - Transport mode (STDIO or HTTP)
OAUTH_PROVIDER - OAuth provider (okta, google, azure)
OAUTH_ENABLED - Enable OAuth authentication
OAUTH_MODE - OAuth mode (native or proxy)
OIDC_ISSUER - OIDC issuer URL
OIDC_AUDIENCE - OIDC audience
OIDC_CLIENT_ID - OAuth client ID

Notes

This project uses oauth-mcp-proxy - a standalone OAuth 2.1 library for Go MCP servers. For production deployment, see the Deployment Guide and OAuth architecture documentation. The server supports multiple data sources through Trino, including PostgreSQL, MySQL, S3/Hive, BigQuery, and MongoDB.

Discussion