What is Open Knowledge Format?

The short definition

OKF describes knowledge as a directory of Markdown files. Each file represents a concept, such as a data table, API endpoint, product rule, support playbook, documentation topic, or website page.

The YAML frontmatter at the top of a file gives machines a few stable fields to scan. The Markdown body gives humans and AI agents the context they need to understand the concept.

What an OKF document contains

• `type`: the concept category. In the draft specification, this is the required field.
• `title`: a human-readable display name.
• `description`: a short summary for previews, index files, and search snippets.
• `resource`: a stable URL, URN, or asset identifier when the concept maps to a resource.
• `tags`: short labels that help group concepts across folders.
• `timestamp`: an ISO 8601 datetime for the last meaningful update.

Why OKF is useful for agents

An agent does not always need a database, embedding pipeline, or custom SDK to understand a knowledge base. If the knowledge is already organized into files, frontmatter, headings, and links, the agent can inspect the structure before loading more content.

That makes OKF useful for teams that want portable context. A bundle can live in Git, ship as a zip archive, or be hosted as static files.

What OKF is not

• OKF is not a Google product page or an official Google service.
• OKF is not a replacement for MCP, RAG, OpenAPI, or llms.txt.
• OKF is not a database. It is a file format for representing curated knowledge.

Who should care