Marie Haynes Demonstrates Building an OKF Brain

For teams designing agent-facing knowledge stores, adopting a standardized file format like OKF reduces integration friction and makes local knowledge more useful to off-the-shelf agents. Marie Haynes' walkthrough synthesizes practical details practitioners can reuse: naming conventions, frontmatter patterns, and an index-driven access model that scopes retrieval for agent workflows.

Marie Haynes, in a June 26, 2026 post on MarieHaynes.com and a republished version on Search Engine Journal, documents building a personal "OKF brain" and publishes a video walkthrough demonstrating the setup. The post shows that every OKF file begins with YAML frontmatter, uses a top-level index.md to enumerate accessible areas, and classifies pages into types such as concepts, entities, playbooks, references, and systems, per the examples in the article.

The post highlights two practical patterns relevant to practitioners. First, using a index.md to narrow an agent's retrieval scope reduces unnecessary RAG computation and focuses context windows. Second, treating each node as a markdown file with typed metadata enables simple graph construction and programmatic linking when new content is ingested. These are general engineering patterns for agentic knowledge bases, not claims about broader OKF governance.

Haynes shows sample folder structure and a snippet of markdown code for a concept page. She describes that, in her system, ingested items such as blog posts or documentation are connected to existing concepts so the resulting structure resembles a knowledge graph; the article includes screenshots illustrating those connections.

Industry observers note that standardized, filesystem-first formats lower the bar for agent experiments because they are human-editable, versionable, and integrate with static-site tooling. Adopting a simple standard also eases sharing knowledge artifacts between tools and teams.

For practitioners building agent pipelines, monitor tooling that automates OKF index generation, connectors that translate other knowledge formats into OKF, and community conventions for metadata fields. Also watch for broader OKF adoption signals from major tooling providers and documentation examples that expand metadata vocabularies.