Open Knowledge Format
Open Knowledge Format packages company metadata and knowledge as markdown files with YAML frontmatter, so any AI agent can read it without a proprietary SDK.
What is the Open Knowledge Format?
The Open Knowledge Format is a way to write down company knowledge, tables, datasets, metrics, runbooks, as markdown files. Google Cloud publishes it. Each file has a block of YAML frontmatter, type, title, description, resource, tags, at the top and a plain explanation underneath, and only type is required. Company knowledge is usually split across a metadata catalog, a wiki, and code comments, each in its own format. OKF gives every source one shape to export into, so any consumer can read it without a dedicated SDK.
How is OKF different from a context repository?
OKF is a file format. A context repository is a system that keeps files like that current. OKF defines no update mechanism, so after export, a file has no way to stay in line with its source once that source changes. A context repository does: a data sync brings each file back in line with its source, resolves the same entity across sources into one record, and applies permissions at retrieval. Qontext’s repository uses the same markdown-plus-frontmatter shape, but adds continuous sync, entity resolution, and permissions on top.
How does OKF work?
The producer, whether that’s an export script, an LLM, or a person, writes one markdown file per concept: a file for each table, dataset, or metric, with links between files that reference each other. Each file’s frontmatter contains fields like type and tags. The body carries the explanation. The bundle contains only files, so it commits to git and renders on GitHub without extra tooling.