azure-aml-schedule-write

Principal can create or update job schedules that periodically trigger pipeline or training jobs on AML compute.

explicit azure emits CanSchedule

match (effective permission)

{ "action": "Microsoft.MachineLearningServices/workspaces/schedules/write", "scope_type": "resource|resourceGroup|subscription" }

emit

source typeIdentity
target typeScheduler
source<principal>
target<AML job schedule>
permissionsMicrosoft.MachineLearningServices/workspaces/schedules/write

Narrative

{principal.name} can create or update AML job schedules (schedules/write) that repeatedly trigger training or pipeline jobs on compute as its managed identity.

Raw rule rules/explicit/azure-machinelearning.yaml

id: azure-aml-schedule-write
emits: CanSchedule
applies_to:
- azure
description: Principal can create or update job schedules that periodically trigger pipeline or training
  jobs on AML compute.
match_effective_permission:
  action: Microsoft.MachineLearningServices/workspaces/schedules/write
  scope_type: resource|resourceGroup|subscription
emit:
  source_type: Identity
  target_type: Scheduler
  source: <principal>
  target: <AML job schedule>
  permissions:
  - Microsoft.MachineLearningServices/workspaces/schedules/write
  api_source: role assignment evaluation
  evidence_field: roleAssignments.properties.roleDefinitionId
  note: Creates a cron/recurrence schedule that triggers a pipeline or command job; persistence + trigger
    creation for repeated code execution as the compute MI.
  false_positive_note: CanSchedule here is persistence / trigger creation only; onward CanExecuteAs still
    requires the referenced compute to have a running (or auto-scalable) cluster and a MI. If the target
    compute has no MI, scheduling provides code execution but not identity escalation.
  narrative: '{principal.name} can create or update AML job schedules (schedules/write) that repeatedly
    trigger training or pipeline jobs on compute as its managed identity.'
move · open · esc close