The OutTheBox governance ontology
Every pattern in this library was written assuming certain other patterns already have an answer: an agent's risk tier, its operating mode, its owner, its cost baseline, whether its model version has changed. This ontology is that structure made explicit and machine-readable, ninety-two patterns, fourteen categories, and the dependencies between them, plus the shared registry schema every pattern reads from or writes to.
It's free, and it's a graph, not a summary: an agent, orchestration layer, or governance tool can query it directly to work out what applies to a given AI agent and in what order, without a human tracing the logic across twenty separate pack downloads by hand.
This is the map: which patterns exist, how they relate, and the shape of the data they operate on. It is not the territory: the decision logic that makes each pattern usable runs inside the governance service. The ontology describes the library; it doesn't replace it.
Published as JSON-LD, valid JSON if you just want to read fields, a real linked-data vocabulary if your tooling wants more. Version 1.1.0, versioned at a stable URL so anything built against it keeps working: earlier field names are kept as deprecated aliases, and the changelog explains every change. Licensed CC BY 4.0, free for any use, including automated or agentic consumption, attribution appreciated.