This MCP server provides a Go-based MCP front-end for Trino, enabling AI assistants and other MCP clients to access Trino's distributed SQL engine through standardized MCP tools. It supports dual transports (HTTP at /mcp and STDIO) and includes OAuth 2.1 authentication via the oauth-mcp-proxy library, offering native mode (client handles OAuth directly) and proxy mode (server proxies OAuth flow). It integrates with four providers (HMAC, Okta, Google, Azure AD) and features token caching, PKCE, and defense-in-depth security. The server exposes MCP tools for common data operations, including execute_query, list_catalogs, list_schemas, list_tables, get_table_schema, and explain_query, enabling seamless query execution and metadata discovery against a Trino cluster. The Data Layer comprises a Trino cluster connected to multiple data sources (PostgreSQL, MySQL, S3/Hive, BigQuery, MongoDB). Docker container support is provided for easy deployment. The project emphasizes backward compatibility with SSE endpoints, performance optimization via schema whitelisting, and broad client compatibility with Cursor, Claude Desktop, Windsurf, ChatWise, and other MCP clients. Deployment guides cover native vs proxy OAuth modes, allowlists, and general configuration.