Evidence-backed facts
- Databricks documents managed MCP servers that expose governed data and AI capabilities as tools for compatible agents and clients. informative
- An MCP client discovers a tool and sends a call to the managed server, which executes within the configured Databricks service boundary. informative
- Workspace identity, Unity Catalog or service permissions, OAuth, endpoint access, tool scopes, and data governance remain independently enforced. informative
- Tool discovery or schema retrieval does not prove permission, execution, result correctness, production use, or that a Databricks workspace is connected. informative