Everything that explains how this works.

A sample world you can take apart, the documentation and API you build against, and the research and evidence behind the compiler.

All resources

Follow a result from an answer back to the exact source location it came from, without signing in.

Representative case — a real compiled result, published read-only.

What a Knowledge Compiler is, and how it differs from a parser, a RAG pipeline and a graph database.

Quickstart, concepts, supported files, compiling, review, and using a world through Ask, the API and MCP.

Endpoints, authentication, errors and limits, with the machine-readable OpenAPI document alongside.

What changed, in the order it changed, written for the people using it.

The open problems in compiling documents into evidence-bound, versioned knowledge.

The eight metric families, the receipt a result must carry, and the rules that decide whether it may be compared.

What an exact source location is for each kind of source, how a result stays bound to one, and how to verify a signed package.

Fixture identity and the material needed to reproduce a published run.

Building something specific? The exact contract is in the documentation.