Triton Inference Server · Agentic Access

Triton Inference Server Agentic Access

x-agentic-access generated

Triton Inference Server exposes 32 API operations that an AI agent could call, of which 14 are state-changing ‘acting’ operations. This is a recommended x-agentic-access execution contract — the scope, audience, consequence tier, short-lived token constraints, and escalation each action should carry before it is handed to an autonomous agent.

By consequence: 18 read and 14 write.

Contracts are classified heuristically from the provider’s OpenAPI and refresh on every APIs.io network build; audience is bound per deployment. The model follows Curity’s Access Intelligence (apidays Munich 2026). Browse every provider’s agent contracts at agentic-access.apis.io.

AIDeep LearningInferenceMachine LearningModel ServingNVIDIAOpen Source
Operations: 32 Acting: 14 Human-in-the-loop: 0 Method: generated

By consequence

read 18 write 14

Source

Agentic Access

Raw ↑ 
generated: '2026-07-15'
method: generated
source: openapi/triton-http-rest-openapi.yml, openapi/triton-metrics-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: 32
  by_action_class:
    connected: 18
    acting: 14
  by_consequence:
    read: 18
    write: 14
  human_in_the_loop_required: 0
operations:
- path: /v2/health/live
  method: get
  operationId: serverLive
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/health/ready
  method: get
  operationId: serverReady
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/versions/{model_version}/ready
  method: get
  operationId: modelVersionReady
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/ready
  method: get
  operationId: modelReady
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2
  method: get
  operationId: serverMetadata
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}
  method: get
  operationId: modelMetadata
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/versions/{model_version}
  method: get
  operationId: modelVersionMetadata
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/config
  method: get
  operationId: modelConfig
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/versions/{model_version}/config
  method: get
  operationId: modelVersionConfig
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/infer
  method: post
  operationId: modelInfer
  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: /v2/models/{model_name}/versions/{model_version}/infer
  method: post
  operationId: modelVersionInfer
  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: /v2/repository/index
  method: post
  operationId: repositoryIndex
  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: /v2/repository/models/{model_name}/load
  method: post
  operationId: modelLoad
  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: /v2/repository/models/{model_name}/unload
  method: post
  operationId: modelUnload
  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: /v2/models/{model_name}/stats
  method: get
  operationId: modelStatistics
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/versions/{model_version}/stats
  method: get
  operationId: modelVersionStatistics
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/stats
  method: get
  operationId: allModelStatistics
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/trace/setting
  method: get
  operationId: getTraceSetting
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/trace/setting
  method: post
  operationId: updateTraceSetting
  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: /v2/models/{model_name}/trace/setting
  method: get
  operationId: getModelTraceSetting
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/models/{model_name}/trace/setting
  method: post
  operationId: updateModelTraceSetting
  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: /v2/logging
  method: get
  operationId: getLogSettings
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/logging
  method: post
  operationId: updateLogSettings
  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: /v2/systemsharedmemory/status
  method: get
  operationId: systemSharedMemoryStatus
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/systemsharedmemory/region/{region_name}/register
  method: post
  operationId: systemSharedMemoryRegister
  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: /v2/systemsharedmemory/region/{region_name}/unregister
  method: post
  operationId: systemSharedMemoryUnregister
  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: /v2/systemsharedmemory/unregister
  method: post
  operationId: systemSharedMemoryUnregisterAll
  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: /v2/cudasharedmemory/status
  method: get
  operationId: cudaSharedMemoryStatus
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none
- path: /v2/cudasharedmemory/region/{region_name}/register
  method: post
  operationId: cudaSharedMemoryRegister
  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: /v2/cudasharedmemory/region/{region_name}/unregister
  method: post
  operationId: cudaSharedMemoryUnregister
  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: /v2/cudasharedmemory/unregister
  method: post
  operationId: cudaSharedMemoryUnregisterAll
  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: /metrics
  method: get
  operationId: getMetrics
  x-agentic-access:
    action-class: connected
    consequence: read
    subject: optional
    token:
      max-ttl: 3600
    audit: none