OKF represents the metadata, context, and curated insight around data and systems as linked Markdown concept files with YAML frontmatter. Human-readable, agent-parseable, zero platform lock‑in.
v0.1
Format version
7
Copy-ready examples
3
Hard requirements
metrics/monthly-recurring-revenue.md
OKF Concept File
Core Concepts
OKF is intentionally minimal. Everything builds on bundles, concepts, and links.
A bundle is a folder you distribute – Git repo, ZIP, or tarball. An index.md maps the contents, a log.md records changes.
One Markdown file per concept. YAML frontmatter with at least type. The file path (minus .md) becomes the concept ID.
Standard Markdown links assert relationships. The prose around the link explains what kind. A # Citations section tracks external sources.
Interactive Explorer
Click files in the tree to inspect concept files. Linked concepts are highlighted.
index.md
markdown
Use Cases
OKF fits curated knowledge that should be read by both humans and agents.
Metrics
Metrics, source tables, dashboards, and review playbooks as linked concepts.
APIs
Endpoints, schemas, errors, and auth without replacing OpenAPI.
Framework
Routes, models, policies, jobs, and operational context for humans and agents.
CMS
Custom post types, taxonomies, ACF fields, and templates as portable knowledge.
Data
Datasets, tables, metrics, lineage, and data quality notes near the systems they describe.
AI
Give agents durable context about systems, tools, playbooks, and constraints before they act.
Parseability
Humans get readable documents. Agents get structured JSON.
Human View
Markdown Rendered
Metric • [revenue, saas, finance]
Agent View
JSON Parsed
Ecosystem
No specialized SDK required. Use the Markdown libraries you already know.
Python
frontmatter package
JavaScript / Node
gray-matter package
Live Playground
Edit YAML frontmatter and Markdown. See the parsed result and validation status in real time.
Start typing or load a preset…
Comparison
OKF doesn't replace these tools – it fills the knowledge layer gap between them.
| Feature | OKF | RAG Chunks | OpenAPI | JSON-LD | Wiki / Docs |
|---|---|---|---|---|---|
| Human-readable | ✓ | ✗ | △ | ✗ | ✓ |
| Agent-parseable | ✓ | △ | ✓ | ✓ | ✗ |
| Structured metadata | ✓ | ✗ | ✓ | ✓ | ✗ |
| Cross-concept links | ✓ | ✗ | △ | ✓ | △ |
| Git-friendly | ✓ | △ | ✓ | ✓ | ✗ |
| No server required | ✓ | ✗ | ✓ | ✓ | ✗ |
| Domain-agnostic | ✓ | ✓ | ✗ | ✓ | ✓ |
| Portable (ZIP / Git) | ✓ | ✗ | ✓ | △ | ✗ |
FAQ
resource URIs and tags.gray-matter for JS, or python-frontmatter) will parse an OKF file perfectly.Getting Started
No SDK, no platform. Just folders, frontmatter, and Markdown.
Add an okf/ directory in your repo, project, or export location.
List your concept groups so humans and agents can scan the bundle at a glance.
One Markdown file per concept. Give it YAML frontmatter with at least a type.
Enrich with description, resource, tags, and timestamp when useful.
Use standard Markdown links. Add # Citations for external sources.
Commit the folder, ship as an archive, or expose it to an agent or catalog.
Help shape the standard, build parsers, or share your knowledge bundles. We're building a common language for humans and AI.
Star on GitHub