This Snowflake MCP server provides tooling for Snowflake Cortex AI features, bringing Cortex capabilities to the MCP ecosystem. When connected to an MCP Client (e.g. Claude for Desktop, fast-agent, Agentic Orchestration Framework), users can leverage Cortex AI features for data querying and orchestration. The MCP server supports Cortex Search (unstructured data queries for Retrieval Augmented Generation), Cortex Analyst (structured data semantic modeling), and Cortex Agent (agentic orchestration across data types). It runs as a sidecar/service and communicates with MCP clients via stdio or sockets (not HTTP). Tools are exposed via server.tool('name', handler), enabling dynamic extension of capabilities such as SQL execution or search. The server uses the Snowflake Cortex REST API for authentication and connections, and supports Programmtic Access Token via environment variables. A Payload Builder constructs Cortex Agent requests dynamically, avoiding hard-coded tool definitions and enabling runtime resource configuration. Responses stream via Server-Sent Events (SSE) and are parsed with processSSEResponse. This separation clarifies that MCP servers are not Next.js API routes. Troubleshooting is aided by the MCP Inspector, which can be launched to diagnose issues.