Google’s Open Knowledge Format (OKF) v0.1: A Vendor‑Neutral AI Standard
Create an OKF bundle with markdown files, YAML frontmatter (type, title, description, resource, tags, timestamp), and publish it as a tarball or Git repo to make content machine‑readable for AI agents.
Build a minimal OKF bundle for your knowledge base, validate that AI crawlers can read it, and monitor Semrush Site Audit for missing llms.txt files.
Summary
Google’s OKF v0.1 is an open specification that packages content as a directory of plain markdown files with YAML frontmatter, making it readable by both humans and AI agents without new runtimes or SDKs. The frontmatter includes fields such as type, title, description, resource, tags, and timestamp, and the bundle can be distributed as a tarball, Git repository, or mounted filesystem. OKF is vendor‑neutral, licensed under Apache 2.0, and currently has reference implementations on Google Cloud, but it is not a search ranking signal. The format is designed to allow AI agents to ingest knowledge directly, unlike sitemaps or llms.txt which direct crawlers outward. While OKF is early and not yet widely adopted, it could become a foundational standard for AI agents if it gains traction similar to schema.org.
Marketers should ensure their content is clean and well‑structured, monitor Semrush Site Audit for missing llms.txt files, and watch how the standard evolves as agents learn what knowledge representations they need.
Key changes
- OKF v0.1 uses plain markdown files with YAML frontmatter
- Fields include type, title, description, resource, tags, timestamp
- Bundle can be a tarball, Git repo, or mounted filesystem
- No compression, runtime, or SDK required
- Spec is Apache 2.0 licensed and vendor‑neutral
- Reference implementations focus on Google Cloud
- OKF is designed for AI agents, not search ranking signals
- Future evolution will adapt to agent knowledge needs