id: role-model-mixed
prompt: "Obtain a reviewer-role assessment first, then use a separate subagent with the explicit model $EVAL_MODEL and no tools to synthesize the final text."
maxCost: 0.1
expectations:
  workflowCallCount:
    min: 1
  minimumAgentCalls: 2
  agentPolicies:
    - callIndex: 0
      role: reviewer
    - callIndex: 1
      model: "$EVAL_MODEL"
      tools:
        mode: empty
semanticCriteria:
  - id: intent
    description: "The reviewer assessment is produced first and passed to a separate no-tools synthesis agent."