The Evolution of LLM Wikis and Open Knowledge Format

The Open Knowledge Format (OKF) is an open-standard specification introduced by Google Cloud to resolve the "context-assembly problem" in AI development. By representing organizational knowledge as a directory of Markdown files with YAML formatter, OKF provides a universal language that allows AI agents to read, write, and interlink internal data without being locked into proprietary platforms. This format is heavily inspired by Andrej Karpathy’s "LLM Wiki" concept, which advocates for AI agents to incrementally build and maintain a persistent, structured knowledge base rather than repeatedly searching raw documents. Unlike traditional retrieval methods, this approach creates a living wiki where information compounds over time, utilizing standardized conventions like "index.md" for navigation and "log.md" for version history. The sources emphasize that OKF is a vendor-neutral format rather than a service, making it highly portable, version-controlled via Git, and accessible to both humans and machines. Practical applications range from automated data cataloging in BigQuery to the creation of self-updating runbooks for engineering teams. Ultimately, the documentation positions OKF as a foundational tool for the Agentic AI era, ensuring that vital institutional knowledge remains interoperable and secure.