UAIX now publishes the Agent Communication Operating Model as a public, machine-readable multi-agent communication standard. The release adds a human guide, eight UAI-1-compatible agent communication profiles, strict JSON schemas, safe valid fixtures, validator policy checks, conformance cases, and a Multi-Agent Communication preset for the AI 记忆数据包向导.
What changed
- Agent Communication Operating Model explains the canonical envelope, source and target identity, correlation IDs, idempotency, delivery fallback, acknowledgement, task status, blockers, memory proposals, handoff, final report, correction, capability negotiation, validation evidence, and support boundaries.
- 模式, 注册表, and 示例 now publish the eight
uai.agent.*.v1profiles and canonical-envelope fixtures. - 验证器 and 一致性包 now include regression coverage for identity, UTC timestamps, expiry, correlation, idempotency, fallback directives, unsupported runtime claims, secrets, final reports, blocker review, and cold-memory promotion.
- The AI 记忆数据包向导 can generate portable multi-agent communication files without becoming hosted messaging, automatic sync, or runtime orchestration.
Support boundary
Agent runtimes execute. UAIX records the reviewed communication, memory, trust, evidence, and handoff boundary. UAIX does not provide automatic sync, hosted messaging, runtime orchestration, official adapters, repository write execution, certification authority, or hosted import mechanisms.
Carcinus clarity
Carcinus.org is a separate runtime/orchestrator example that may consume UAIX records. UAIX.org does not implement Carcinus runtime behavior and does not execute Carcinus workflows.
LocalEndPoint clarity
LocalEndPoint and other runtimes may consume the published UAIX records as external consumers; UAIX.org does not implement their project-specific runtime features.