generated: '2026-09-07'
method: generated+corrected
source: openapi/advanceai-openapi.yml
description: Recommended x-agentic-access execution contracts, classified heuristically from the OpenAPI.
  A governance starting point for exposing this API to AI agents — review and bind audience per deployment.
  See research/curity/agentic-governance/.
summary:
  operations: 9
  by_action_class:
    acting: 8
    connected: 1
  by_consequence:
    write: 7
    read: 1
    destructive: 1
  human_in_the_loop_required: 1
operations:
- path: /openapi/auth/ticket/v1/generate-token
  method: post
  operationId: generateAccessToken
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /intl/openapi/face-identity/document-verification/v1/auth-license
  method: post
  operationId: authorizeDocumentVerificationLicense
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /intl/openapi/face-identity/document-verification/v1/query
  method: post
  operationId: queryDocumentVerificationResult
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /openapi/face-recognition/v4/check
  method: post
  operationId: compareFaces
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /liveness/ext/v1/generate-signature-id
  method: post
  operationId: generateLivenessSignatureId
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /openapi/liveness/v1/auth-license
  method: post
  operationId: authorizeLivenessLicense
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /openapi/liveness/v3/detection-result
  method: post
  operationId: getLivenessDetectionResult
  x-agentic-access:
    action-class: acting
    consequence: write
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: conditional
      triggers:
      - abnormal
      - high-value
    audit: required
- path: /liveness/ext/v1/get-video
  method: get
  operationId: getLivenessVideo
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: required
    token:
      max-ttl: 900
    audit: required
    data-sensitivity: biometric
    note: Returns a link to biometric video of an identified person. Reads of this class should be audited.
- path: /liveness/ext/v1/clear-data
  method: get
  operationId: clearLivenessPiiData
  x-agentic-access:
    action-class: acting
    consequence: destructive
    reversible: false
    subject: required
    audience: null
    token:
      max-ttl: 900
    escalation:
      human-in-the-loop: required
      triggers:
      - always
      reason: Irreversible. ADVANCE.AI documents no restore, no soft delete and no recovery window.
    audit: required
    note: Implemented as GET but deletes data. See conventions/advanceai-conventions.yml reversibility.
x-manual-correction:
  date: '2026-09-07'
  by: API Evangelist enrichment pass
  reason: The heuristic classifies by HTTP method, and ADVANCE.AI implements two consequential operations
    as GET. clearLivenessPiiData permanently deletes the PII held for a liveness detection and was auto-classified
    consequence:read / audit:none / no escalation — the most dangerous possible misreading, since it would
    tell an agent an irreversible deletion is a safe read. getLivenessVideo returns biometric video and
    was likewise classified as an unaudited read. Both entries below were corrected by hand from https://doc.advance.ai/liveness_detection.html.
  corrected_operations:
  - clearLivenessPiiData
  - getLivenessVideo
  caveat: A GET that mutates is invisible to method-based classification. Any other provider in this catalog
    that deletes over GET carries the same silent defect.