LoginStart beta
*Ayneye
MCP quickstartStart beta

MCP policy contract

Define what the agent may say before it calls a video tool.

The policy contract is the guardrail between evidence retrieval and product behavior. It makes allowed claims, blocked claims, review triggers, and audit requirements explicit.

MCP policy contract

{{
  "allowed_claims": ["object_presence", "event_time", "zone_entry", "supported_summary"],
  "blocked_claims": ["identity", "intent", "guilt", "medical", "legal", "employment_action"],
  "blocked_actions": ["unlock_door", "dispatch_emergency", "contact_authority", "delete_video"],
  "requires_evidence_refs": true,
  "requires_review_when": ["low_confidence", "security_outcome", "physical_access", "person_related"],
  "fallback_response": "insufficient_evidence"
}}

Policy lifecycle

StageWhat changesWho reviews
BetaRead-only artifact access, evidence-required answersProduct and engineering
PilotExpanded quotas, more videos, stricter auditSecurity/legal/ops
ProductionCustomer-specific policies, retention, monitoring, access rolesFormal governance owner

Audit requirement

Every tool call should record tenant, user, video_id, artifact version, tool name, question, evidence refs, blocked operations, review_required state, and cost context. Without that trace, a video-agent workflow cannot be debugged or defended during procurement.

The purpose of this page is to make the MCP boundary concrete for builders, buyers, and reviewers. A useful MCP integration is not a promise that an agent can watch everything and decide everything. It is a disciplined interface over already-materialized evidence. Each tool call should be tenant-scoped, read-only by default, evidence-aware, cost-aware, and auditable. If the request asks for identity, intent, physical access, emergency response, employment action, or destructive change, the tool returns a review-required state instead of pretending that a video artifact is a final decision. This keeps the product useful for developers while making it credible for security, legal, and operations teams.