Your business needs an AI platform where users get governed answers from live data, knowledge workers search institutional memory, and agents orchestrate complex workflows, all while partners access the same capabilities through their own identity providers.
The tools will change. The principles won't. Govern the invariants.
This repository is a gold standard for identity, data governance, tool and API governance, observability, and agent runtime governance on Databricks. Reference documentation covers authentication, authorization, federation, fine-grained access control, custom MCP and tool policy, audit, and the policy-guardrail harness. The presentation library covers IAM, AI Gateway, cost control, UC governance, and cross-IdP federation.
Entry point: the overall governance model, with links to every deeper deck below.
One IdP layer, U2M/OBO/M2M token paths, auth vs authz, decision guide.
Auth mechanics per resource: serving endpoints, Genie Agents, Vector Search, UC Functions, Lakebase.
Two-leg token exchange, per-user OBO vs. role-based SP, federation policy config, implementation checklist.
Four-layer access control, row filters, column masks, ABAC.
Agent Bricks, MCP tool governance, agent auth methods, UC Connections as the governance primitive.
Identity, runtime, and observability for LLM/MCP/API traffic. Includes Databricks vs. external gateway vs. UC-native controls.