Jinja-Based Arbitration

Runs Jinja arbitration parsing consensus vs disagreement.

name: Jinja-Based Arbitration
version: "1.0.0"
description: Runs Jinja arbitration parsing consensus vs disagreement.
metadata:
  domain: clinical
  complexity: medium
  tags:
    - clinical
    - arbitration
variables:
  - name: consensus_reached
    description: Boolean or string indicating if consensus exists.
    required: true
  - name: meta_orchestration_summary
    description: Consolidated summary from meta-orchestrator.
    required: true
model: gpt-4o-mini
modelParameters:
  temperature: 0.1
messages:
  - role: system
    content: |
      You are an automated clinical review arbitrator.
  - role: user
    content: |
      Consensus Reached: {{consensus_reached}}
      Summary: {{meta_orchestration_summary}}
testData:
  - inputs:
      consensus_reached: "True"
    expected: "consensus"
  - inputs:
      consensus_reached: "False"
    expected: "disagreement"
evaluators: []