Google Cloud Open Knowledge Format launches as Markdown standard

Google Cloud launched Open Knowledge Format (OKF) this week, a specification that standardizes organizational knowledge as Markdown files with YAML frontmatter. The format aims to make documents, notes and wiki pages portable so AI agents, ingestion pipelines and search systems can consume them consistently.

OKF sets a simple file-based convention rather than a proprietary database or locked system. Files are plain Markdown with a small YAML header carrying structured metadata such as title, summary, authorship, tags and canonical source. The specification emphasizes human-readability, Git-friendly storage and predictable metadata fields so tools can index and reason over content without custom connectors.

What OKF defines

The core of OKF is twofold: content and metadata. Content is the document body in Markdown. Metadata lives in YAML frontmatter at the top of the file and signals the record type and key fields. Examples include a document type label, a short description, canonical URL, creation and modification timestamps, and author identifiers. The spec also outlines optional fields for attachments, redirects and access control hints.

Google Cloud positions OKF as a bridge between scattered knowledge stores and the LLM-driven tooling that consumes that knowledge. The format does not mandate a specific schema beyond a small set of recommended fields. That gives organizations leeway to include enterprise-specific metadata while preserving enough consistency for generic tools to operate.

The specification includes examples for common use cases: converting wiki pages into OKF Markdown, exporting Google Docs or other text artifacts as OKF records, and composing OKF records that reference images and binary attachments. Google Cloud provided examples that show how links and anchors should be represented so downstream agents can follow references and resolve content relationships.

How customers and tools can use it

OKF is intended to be file-centric and compatible with existing developer workflows. Organizations can store OKF files in Git repositories, cloud storage buckets, or within content management systems that support static assets. That file-centric approach aims to simplify versioning, review and collaboration.

Tool vendors and internal teams can adopt OKF in two main ways. First, ingest: converters turn source artifacts (docs, wikis, notes) into OKF Markdown, normalizing metadata so an indexer or vectorizer sees consistent fields. Second, runtime use: AI agents and retrieval systems read OKF files directly, relying on the YAML frontmatter for routing, filtering or display hints before passing the Markdown body to an LLM.

Google Cloud described OKF as an interoperability layer: it reduces the need for bespoke connectors for every content platform. The format can also sit alongside existing enterprise metadata systems; OKF files can carry pointers back to canonical records in a primary content system while still being usable as a portable snapshot.

Early adopters are likely to be engineering and data teams building agent frameworks, search infrastructure teams aiming to consolidate indexes, and tooling vendors that provide document ingestion and normalization. Vendors that publish OKF exporters or CLI utilities will make it easier for organizations to convert legacy content stores into the format.

Why it matters

A simple, file-based specification lowers friction for making organizational knowledge available to AI agents and indexing tools. By standardizing metadata and using widely supported formats, OKF could reduce the number of custom integrations teams must build to feed LLMs. If adopted broadly, the format could make content portability and auditability easier across tools and vendors.