Open Knowledge FormatBuild an AI-maintained knowledge base — from beginner to enterprise
A knowledge base stored as plain Markdown files that humans and AI agents can read, write, and use together.
Preface
Welcome to The Open Knowledge Format (OKF) Handbook. This book walks you through building and operating an AI-maintained knowledge base with OKF — a knowledge base stored as Markdown files with YAML frontmatter that both humans and AI agents can use, without any SDK or special database.
🌐 อ่านฉบับภาษาไทยได้ที่ปุ่ม ไทย มุมขวาบน (Thai version available — click ไทย in the top bar)
Table of Contents
Who this book is for
- Beginners who want a searchable, vendor-neutral personal/team knowledge base
- Developers / data teams who want AI agents to reach internal knowledge systematically
- Architects / platform teams who need a shared system across sessions and teams, on-prem
No prior OKF knowledge needed — just basic command line and Git.
Conventions
- Commands you type go in code blocks · technical terms (
concept,frontmatter,bundle,MCP) stay in English to match the source docs and code · blockquotes are warnings/tips
Versioning note: OKF is a v0.1 spec (published 2026-06-12 by Google Cloud). The core spec requires only the
typefield — most "best practices" here come from the LLM-wiki community and Google's reference implementation.
Source project (code + all tools): https://github.com/supachai-j/open-knowledge-format-starter
Start at What is OKF