OpenMOCA is an open standard for portable grounding for AI. MOCA stands for
Modular Ontology, Content & Attestation.
|
|
| Why MOCA? |
The problem, what MOCA adds to OKF, and what it is not |
| MOCA and RAG |
What a package adds to retrieval-augmented generation |
| Trust and provenance |
Who published it, who checked it, whether it is current, and proof of what an answer used |
| Grounding for AI agents |
One package for any agent: MCP, framework bindings and in-process Readers |
| Ontology for AI retrieval |
Topics, order and prerequisites as structure, and retrieval that uses them |
| Sovereign and offline AI |
Grounding that runs on infrastructure you control, with no outside calls |
| Use cases |
Where a package layer pays off, and where it does not |
| Architecture |
MOCA Package, Reader and Application; layers and overlays; how it fits your retrieval stack |
| MOCA, OKF and MCP |
How the three fit together |
| FAQ |
Short answers: what MOCA is, and how it relates to RAG, vector stores, OKF and MCP |
|
|
| Quickstart |
A valid package in five minutes |
| Walkthrough |
Check, sign, review, index and serve a package to an AI agent |
| Authoring packages |
Content, evidence, validity, versions, members, relations |
| Structure |
structure.ttl, concept bindings, structure operations, overlays, retrieval strategies |
| Signing and review |
Publisher and review attestations, trust roots |
| Consuming packages |
Use a Reader in TypeScript, .NET or Python: read and verify, search, citation records, your own store, MCP |
| Search and sidecar indexes |
One search, backends, scope, hooks, sidecars, vector stores |
| Capabilities |
Which features your use needs |
The normative documents are in spec/.
