An open protocol in working draft

Project purpose

The project exists to make useful AI context portable without creating a universal dossier or a hidden authorization channel. Its design keeps people, consent, provenance, loss, custody, and review state visible wherever context moves.

What OCP does not own

• Authentication, identity proofing, credential issuance, or account recovery.
• Business-domain authority, agent permissions, payments, or external actions.
• A universal ontology for every personal, organizational, or model-memory concept.
• Model behavior, internal reasoning, safety policy, or a provider’s storage architecture.
• Automatic truth decisions or silent promotion of imported material into accepted state.

Governance and license status

Neutral stewardship, contribution and IPR policy, patent policy, trademark rules, conformance marks, registry administration, vulnerability response, and final licenses have not yet been adopted.

How to evaluate the draft

1. Read the boundaries.Start with the introduction, core concepts, and safety guidance.
2. Run the release suite.Reproduce the local structural and semantic checks without relabeling them as independent conformance.
3. Build a narrow implementation.Name the system under test, conformance class, build, fixtures, and results.
4. Report exact gaps.Separate editorial issues, implementation behavior, security evidence, and interoperability evidence.

Deferred profile work

Organization authority, minors and guardianship, inheritance, full multi-party context, autonomous delegation, live delta synchronization, offline CRDT merge, universal credential or personal-context semantics, universal tool history, and jurisdictional profiles remain future work unless implementation evidence exposes a required narrow hook.