Google Unveils Open Knowledge Format - A Knowledge-Sharing Standard for AI Agents

Google has released the Open Knowledge Format (OKF) to solve the context fragmentation problem that arises when building AI agent systems. Previously, knowledge within organizations was scattered across multiple systems such as wikis, code comments, and databases, forcing agents to reassemble context every time.

OKF v0.1 is a vendor-neutral open specification that is not tied to any specific platform. It represents knowledge as a directory of Markdown files with YAML frontmatter, and is designed so that anyone can produce and consume it without complex SDKs or new runtimes.

Key Features and Design Principles:

• Minimal rules: Aside from the type field, no format is enforced, allowing flexible content modeling.
• Producer/consumer independence: Agents can read human-written documents, and other LLMs can immediately use data generated by automated pipelines.
• Platform independence: Not tied to any specific cloud or tool, and can be version-controlled alongside code (e.g., Git).
• Graph structure support: Supports cross-referencing between files via Markdown links, forming a rich knowledge graph beyond a simple directory structure.

To promote adoption of the format, Google is also providing reference implementations, including an Enrichment agent that automatically generates OKF documents from BigQuery datasets, and a static HTML-based visualizer.