Knowledge Base

The Knowledge Base renders the published ontology as a human- and agent-readable wiki. It runs automatically on a configurable cadence — for example daily, or every few hours — compiling all of your organization's ontology resources into a single maintained knowledge bundle.

The Knowledge Base is a rendering of the ontology, not an independent source of truth. It does not author knowledge of its own; everything it contains is derived from published primitives. Each scheduled run regenerates and gardens the bundle so that it reflects the current state of the ontology, including any updates the Ontology Agent has published since the previous run.

Output format

The Knowledge Base emits its wiki in the Open Knowledge Format (OKF), an open, vendor-neutral specification published by Google Cloud in June 2026 for representing knowledge as a directory of markdown files with YAML frontmatter.

OKF is deliberately minimal:

• Each concept — a thing type, a metric, a process — is a single markdown file.
• Files interlink with ordinary markdown links, forming a navigable graph.
• Frontmatter carries metadata, with type as the only required field.
• By convention, an index.md enumerates the contents of a directory to support progressive disclosure, and a log.md records changes over time as a changelog that travels with the knowledge — a natural destination for the updates surfaced by drift detection.

Because the output is plain markdown and YAML, it:

• requires no runtime or SDK to read,
• renders directly in any tool that displays markdown,
• is diffable in version control, and
• can be consumed by any agent or system that understands the format.

Choosing OKF as the output format keeps your knowledge portable: structured enough to be useful, open enough to avoid lock-in.

Relationship to the Ontology Agent

The Ontology Agent maintains the structured model; the Knowledge Base presents it. The two are not separate agents but two faces of the same maintained knowledge — one optimized for systems and modeling, the other for reading and interoperability.