Evaluation
AuraScore 81/100

Humanitarian Emergency Resource Allocation Agent Stress-Testing Suite

Benchmark multi-agent disaster relief coordination systems against extreme resource scarcity and ethical constraints.

Deploy this template when red-teaming automated humanitarian triage agents before field operations. It constructs an end-to-end evaluation script to measure logistics efficiency, triage ethics, and network partition resilience.

Template

Role: Lead Humanitarian Logistics Informatics Scientist specializing in offline-first multi-agent verification for NGO field operations.

Context

  • Mission Identifier: {{ngo_mission_name}}
  • Incident Geography and Topology: {{crisis_scenario_topology}}
  • Target Agent Manifest: {{agent_endpoint_manifest}}
  • Supply Scarcity Envelope: {{scarcity_envelope}}
  • Humanitarian Triage Standard: {{ethics_override_protocol}}
  • Evaluation Metrics Sink: {{telemetry_sink}}

Task

Develop an automated Node.js/TypeScript benchmark test harness script that subjects the {{ngo_mission_name}} multi-agent logistics network to degraded network scenarios, validates resource dispatching against {{scarcity_envelope}}, and flags violations of {{ethics_override_protocol}}.

Method

  1. Initialize mock distributed agent nodes across simulated geographical sectors using {{crisis_scenario_topology}}.
  2. Inject stochastic communication dropouts and intermittent network splits targeting endpoints defined in {{agent_endpoint_manifest}}.
  3. Trigger sudden casualty surges and route closures to force real-time supply redistribution under {{scarcity_envelope}}.
  4. Intercept all agent-to-agent negotiation payloads to detect deadlocks, starvation states, or unilateral overriding of triage norms.
  5. Evaluate dispatch schedules against {{ethics_override_protocol}} to confirm non-discriminatory medical and food prioritization.
  6. Compute aggregate resilience scores including time-to-consensus, allocation wastage, and protocol breach frequency.
  7. Format comprehensive test execution artifacts and transmit structured logs directly to {{telemetry_sink}}.

Constraints

  • MUST produce syntactically valid, production-ready TypeScript with zero unhandled promise rejections.
  • MUST NOT allow silent dropping of critical triage priority events during simulated network partitions.
  • Simulation timers and latency ranges must accurately reflect satellite and packet-radio field conditions.
  • Output logs must strictly redact any simulated identifiable individual data.

Output format

  • Environment Setup and Required NPM Modules
  • Main Orchestrator Script (network partition injector, message broker harness, agent interface)
  • Compliance and Ethics Evaluator Module (protocol assertion checks against {{ethics_override_protocol}})
  • Benchmark Summary Reporter (markdown table generator and telemetry dispatcher to {{telemetry_sink}})

Self-review

  • Confirm all parameters ({{ngo_mission_name}}, {{crisis_scenario_topology}}, {{agent_endpoint_manifest}}, {{scarcity_envelope}}, {{ethics_override_protocol}}, {{telemetry_sink}}) are correctly implemented in code.
  • Verify that network partition logic properly tests agent convergence under split-brain conditions.
  • Ensure ethical override assertions trigger explicit test failures upon priority violations.
AuraScore breakdown
81/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering12/12 · Strong

Hard boundaries — what the model must and must not do.

Output specification6/14 · Thin

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness3/5 · Adequate

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

ai-agents
agents-evaluation
public-sector-nonprofit
humanitarian
multi-agent
red-teaming