registryajp:AgentIncident
AJP AI AGENT SCHEMA STANDARD

ajp:AgentIncident

extends schema:Event

A formal record of an incident where an AI agent behaved outside its declared capability boundaries or caused harm. The machine-readable evidence standard for agent disputes, linked to the deploying business, the agent passport, and the resolution outcome.

REQUIRED PROPERTIES (8)
ajp:incidentIdTextrequired

Unique incident identifier

ajp:incidentDateDateTimerequired

When the incident occurred

ajp:incidentTypeTextrequired

prohibited_action_violated / hallucination / data_breach / unauthorised_action / other

ajp:affectedDeploymentajp:AgentDeploymentrequired

The specific deployment where the incident occurred

ajp:affectedAgentajp:AIAgentrequired

The agent involved

ajp:allegedViolationTextrequired

Plain-English description of what went wrong

ajp:evidenceDescriptionTextrequired

What evidence exists

ajp:hashMatchAtIncidentBooleanrequired

Was the agent system prompt hash intact at time of incident?

OPTIONAL PROPERTIES (8)
ajp:allegedProhibitedActionajp:ProhibitedAction

Which specific prohibited action was violated

ajp:evidenceHashText

SHA-256 of the evidence package

ajp:resolutionOutcomeText

outcome_a / outcome_b / outcome_c / pending

ajp:resolutionDateDateTime

When resolved

ajp:resolutionSummaryText

Plain-English resolution

ajp:refundIssuedBoolean

Whether a refund was issued

ajp:builderStrikeIssuedBoolean

Whether builder received a strike

ajp:reviewedByschema:Organization

Who reviewed the incident

EXAMPLE JSON-LD
{
  "@context": {
    "@vocab": "https://schema.org/",
    "ajp": "https://schema.anthonyjamespeacock.com/"
  },
  "@type": [
    "ajp:AgentIncident"
  ],
  "ajp:incidentId": "<Text>",
  "ajp:incidentDate": "<DateTime>",
  "ajp:incidentType": "<Text>"
}
MACHINE ENDPOINT
GET /api/schema/AgentIncident
Accept: application/ld+json
View raw JSON-LD →
OTHER TYPES