Mekra

Mekra takes its name from the Korean word 맥락, meaning context: the background that makes a fact understandable — why it matters, under which conditions it holds, and how it relates to other knowledge.

Mekra is not a folder layout or a document format. It is a way of looking at how knowledge, context, and judgment fit together when people and AI agents work on the same project, and a commitment to keeping that context readable where it is needed.

Mekra grew out of research into and practical work with Open Knowledge Format (OKF). Along the way, the project carried its knowledge and history through the names OKF Lab → OKF Method → Mekra Method.

Keep sources of truth clear, share context where it is needed, and delegate judgment.

The first idea is to keep definition and change responsibility clear while deliberately distributing the context needed for understanding. A fact has a clear place where it is defined and changed, and its effect on the meaning, conditions, and exceptions of other concepts is explained where those concepts are discussed. A single source of truth does not imply a single location for context.

The second idea is to trust capable agents to interpret context and exercise judgment. Instead of prescribing every sequence of work, provide the reasons, conditions, and relationships that support judgment, and leave the concrete way of working to the agent. That autonomy rests on evidence and context the agent can actually find and read — and distributing context brings a responsibility to maintain it together.

Adopted principles

• Distinguish canonical responsibility from the location of context. Repeating a fact is not the problem; defining and changing it independently in several places is. Source of truth and context
• Retain the meaning, reasons, conditions, and relationships needed for judgment. Knowledge internalization
• Reflect the meaning of a change in related knowledge. Context propagation
• Choose structure and depth from actual need. Operating principles
• Start from autonomous agent judgment. Discretion over interpretation and working method is distinct from authority to change facts, policies, or permissions. Agent autonomy
• Distinguish the status of a record from the certainty of its content. Source of truth and context

Ideas still under investigation are developed in Mekra Lab’s research repository. They appear here as principles only once they are adopted.

Mekra Method helps you organize and maintain project knowledge so AI agents can understand its context and carry it into later work. If you keep explaining the same background, or keep tracing how one decision affects another task, you can record those reasons and relationships where they matter.

It is currently built on Open Knowledge Format (OKF), a format for representing knowledge that people and agents can both read and exchange. Mekra’s operating principles are its own judgments on top of that specification, not additional official requirements.

The guide is organized into:

• Application guide — exploring a target, resolving questions, making changes, and continuing to use the result
• Operating knowledge — adopted principles and the reasoning behind them
• Facets — lenses for finding important judgments from a target’s properties
• Templates — minimal scaffolding to copy and adapt
• Versions — the recommended OKF baseline, Mekra Method releases, and migration records

In the repository you want to work on, ask an agent that can read and edit it:

The agent examines the existing material and structure, judges which knowledge to organize and connect, and afterward explains where new material and updates belong.

There is no exact wording to learn, and you can describe the work more specifically.

The method, its application materials, and its change history live on GitHub. This site introduces them; the repositories remain the source of truth.