OKF guide
What Is Open Knowledge Format (OKF)?
Open Knowledge Format (OKF) is an open specification from Google Cloud for packaging curated knowledge as a directory of Markdown files with YAML frontmatter, so both people and AI agents can read it.
The short version
OKF stores knowledge as plain Markdown files with a small block of YAML frontmatter at the top. It is just Markdown, just files, and just a few structured fields: no database, no API, and no SDK required. An AI agent reads or writes a file directly, so the file system itself becomes the interface.
How an OKF file is structured
Each OKF document is Markdown for the human-readable body plus typed frontmatter for the small set of fields that need to be queryable: type, title, description, resource, tags, and timestamp. The body stays free-form Markdown, so it renders anywhere and stays easy to edit.
Why AI agents use OKF
Agents work best with curated, stable context. OKF gives them a pre-compiled knowledge layer they can load directly instead of re-deriving it every time. It complements llms.txt (a site-level map) and the Model Context Protocol (a way to fetch data on request).
Related OKF tools and guides
FAQ
Who created Open Knowledge Format?
OKF was published by the Google Cloud data team as an open, Apache-2.0 specification. Version 0.1 is intentionally small and is expected to evolve as more producers and consumers adopt it.
Is OKF the same as llms.txt?
No. llms.txt is a single site-level index that points AI systems to important URLs. OKF is the curated knowledge itself: a bundle of typed Markdown files agents load into context.
How do I start using OKF?
Start from a template, paste it into the OKF Validator to check the frontmatter, then publish focused files for your APIs, datasets, metrics, and product docs.