Plan Proactive Agent Eval Guardrails
plan_proactive_agent_eval_guardrailsRead-only
Maps evaluation gaps in proactive assistants to guardrails for state-machine, user simulation, goal inference, intervention timing, and multi-app orchestration.
Instructions
Map proactive-assistant eval gaps to PARE-style state-machine, active-user-simulation, goal-inference, intervention-timing, and multi-app orchestration gates.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| workflow | No | Proactive assistant workflow name. | |
| apps | No | Apps involved in the proactive workflow. | |
| states | No | Modeled app states. | |
| stateCount | No | Number of modeled states. | |
| actionCount | No | Number of state-dependent actions. | |
| taskCount | No | Number of benchmark tasks or scenarios. | |
| hasStateMachine | No | Whether apps are modeled as finite state machines. | |
| hasActiveUserSimulation | No | Whether active user simulation exists. | |
| hasGoalInferenceEvals | No | Whether goal inference is graded. | |
| hasInterventionTimingEvals | No | Whether intervention timing is graded. | |
| hasMultiAppEvals | No | Whether multi-app orchestration is graded. | |
| flatToolApiOnly | No | Current eval only covers flat tool calls. | |
| proactiveWrites | No | Proactive agent can write or mutate state. | |
| userVisibleActions | No | Interventions can notify, schedule, send, or affect users. |