Plain-English purpose
Reports a boundary, secret, dependency, validation, scope, access, or other blocker that prevents safe continuation.
When to use it
Use it when the next action needs human review, redaction, new authority, or another explicit unblock decision.
When not to use it
Do not bury blocker material inside a status message when secret or authority concerns are involved.
Required fields
All agent communication packets support and require these canonical envelope fields:
uai_version, profile, message_id, source, target, conversation, delivery, trust, body, provenance, integrity, extensions
The body object for this profile requires: blocker_id, blocker_type, description, human_review_required, proposed_resolution, support_boundary.
Reliability fields
Use conversation.correlation_id, delivery.idempotency_key, delivery.timeout_ms, delivery.retry_count, delivery.fallback_directive, and delivery.expected_output_schema to make delegation reviewable without turning UAIX into a runtime.
Additional fields
No undeclared top-level fields are allowed. Keep additional data inside the published envelope objects and use empty arrays or empty objects where the schema expects them.
Validation expectations
- Schema validation is strict and rejects undeclared fields.
- All date/time fields use UTC with a trailing
Z. - Malformed source or target identity fields fail validation.
- Missing or malformed correlation IDs fail validation.
- Missing idempotency keys or fallback directives fail validation.
- Expired packets fail validation.
- Unsupported claims about UAIX hosted messaging, automatic sync, runtime orchestration, repository writes, official adapters, certification, or hosted import fail validation.
- Memory proposals containing secrets or token-like strings fail validation.
- Cold-memory records cannot be promoted directly into current truth.
- Final reports must include complete evidence, not empty done messages.
- Authorization, secret, destructive-action, or boundary-conflict blockers require human review.
Trust and provenance expectations
Use trust, provenance, integrity, and profile-specific body evidence to preserve who produced the packet, what was reviewed, which schema and validator were expected, and which evidence supports the message. Agent runtimes execute. UAIX records the reviewed communication, memory, trust, evidence, and handoff boundary.
Redaction rules
Do not include private keys, API credentials, session tokens, write tokens, temporary beta instructions, local error traces, or private operational endpoints. Redact before publication and reject memory proposals that contain secret-like strings.
Support boundary
UAIX does not provide automatic sync, hosted messaging, runtime orchestration, official adapters, repository write execution, certification authority, or hosted import mechanisms. 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.
Related UAIX records
Valid JSON fixture example
{
"uai_version": "1.0",
"profile": "uai.agent.blocker.v1",
"message_id": "msg_uaix_agent_blocker_0004",
"source": {
"type": "agent",
"id": "agt_uaix.publisher",
"label": "UAIX Publication Agent",
"uri": "https://uaix.org/",
"role": "standard-publisher",
"project": "uaix.org",
"implementation": "uaix-core"
},
"target": {
"type": "runtime",
"id": "agt_runtime.consumer",
"label": "Runtime Implementer",
"uri": "https://example.invalid/runtime",
"role": "runtime-implementer",
"project": "example-runtime",
"implementation": "external-runtime"
},
"conversation": {
"conversation_id": "conv_20260531_agent_standard",
"turn_id": "turn_004",
"parent_message_id": "msg_uaix_agent_message_0001",
"correlation_id": "corr_20260531_agent_standard",
"traceparent": "00-11111111111111111111111111111111-0000000000000004-01",
"sequence": 4
},
"delivery": {
"mode": "async",
"priority": "interactive",
"expires_at": "2030-12-31T23:59:00Z",
"reply_requested": false,
"ack_required": false,
"task_ref": "task_uaix_agent_standard_001",
"idempotency_key": "idem_20260531_agent_blocker_0004",
"retry_count": 0,
"sequence": 4,
"lifecycle": "blocked",
"timeout_ms": 300000,
"fallback_directive": "Return a uai.agent.blocker.v1 record if the target cannot proceed safely.",
"expected_output_schema": {
"profile": "uai.agent.ack.v1"
}
},
"trust": {
"channel": "public-web",
"auth_scheme": "https",
"principal": "https://uaix.org",
"signature_ref": "https://uaix.org/.well-known/uaix.json",
"replay_window_id": "rw-uaix-agent-standard-20260531"
},
"body": {
"blocker_id": "blk_uaix_authorization_001",
"blocker_type": "authorization",
"description": "The receiver needs human authorization before performing a deployment or publishing action.",
"human_review_required": true,
"proposed_resolution": "Obtain explicit human approval, then send a new idempotent directive packet.",
"support_boundary": "Agent runtimes execute. UAIX records the reviewed communication, memory, trust, evidence, and handoff boundary."
},
"provenance": {
"trace_id": "trace_20260531_agent_standard",
"parent_trace_id": "trace_root_uaix_agent_standard",
"issued_at": "2026-05-31T18:00:00Z",
"log_ref": "urn:uaix:log:agent-standard:20260531:blocker_0004",
"agent_id": "agt_uaix.publisher",
"model_id": "model.uaix.standard-publication",
"confidence": 1,
"lineage": [
{
"stage": "agent-communication-standard-release",
"actor_id": "agt_uaix.publisher",
"model_id": "model.uaix.standard-publication",
"note": "Generated as a safe public fixture for the UAIX Agent Communication Operating Model."
}
]
},
"integrity": {
"version": 1,
"algorithm": "sha256",
"canonicalization": "jcs-compatible-sorted-json",
"checksum": "sha256:fixture-hash-recorded-by-validator"
},
"extensions": [
{
"namespace": "urn:uaix:ext:agent-communication-standard",
"purpose": "Documents UAI-1 agent communication expansion fields without executing runtime work.",
"critical": false
}
]
}