Connects Claude to your Marmot data catalog instance via streamable HTTP, letting the AI search across tables, topics, queues, and APIs cataloged in your system. Useful when you want Claude to answer questions about what data assets exist, trace lineage between datasets, or look up ownership and documentation without leaving the conversation. Since Marmot stores rich metadata and runs as a single binary, this works well for teams who've already deployed it and want their AI to reference the same catalog humans use. Point it at your Marmot instance and Claude can query the same search, lineage, and ownership APIs exposed through the v1 MCP endpoint.