Multi-Agent Policy Impact Analysis
Develop an advanced multi-agent system using AutoGen, powered by GPT-5, to monitor and analyze public policy changes (e.g., government announcements, regulatory updates) and assess their potential impact on specific stakeholders (e.g., journalists, businesses, citizens). This system should identify subtle shifts, interpret complex legal or political language, and generate detailed impact reports or proactive alerts. The challenge requires implementing a hierarchical agent structure with specialized agents for policy monitoring, legal interpretation, stakeholder impact assessment, and report generation. Crucially, the agents must communicate securely using an A2A (Agent-to-Agent) protocol and integrate external tools via an MCP (Multi-Agent Coordination Protocol) for accessing official government databases, news feeds, and social media. Adaptive reasoning budgets should be employed by GPT-5 to allocate more processing power for ambiguous or high-stakes policy changes.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a hierarchical multi-agent system with AutoGen and GPT-5, using A2A protocol and MCP tools for policy impact analysis, leveraging adaptive reasoning budgets.
How work is evaluated
The evaluation will assess the AutoGen system's ability to accurately identify policy changes, interpret their implications, quantify stakeholder impact, and generate high-quality, actionable reports using A2A and MCP tools.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Utilize GPT-5 with extended thinking and adaptive reasoning budgets for nuanced policy interpretation.
Implement a secure A2A Protocol for structured data sharing between agents.
Build hierarchical multi-agent systems with AutoGen for complex analytical tasks.
Integrate MCP-enabled tools for comprehensive data ingestion from diverse sources.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation will assess the AutoGen system's ability to accurately identify policy changes, interpret their implications, quantify stakeholder impact, and generate high-quality, actionable reports using A2A and MCP...
Challenge input
{"news_articles": "list_of_strings", "official_releases": "list_of_strings"}
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{"policy_identified": "boolean", "policy_summary": "string", "core_intent_interpretation": "string"}
- Verify agents exchange information securely and with proper authorization via the A2A protocol.
- Ensure agents successfully interact with external data sources using MCP-enabled tools.
- Assess if the core intent of the policy change is correctly identified and summarized by GPT-5.
- Impact Assessment Granularity target: 0.85
- Python execution harness
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- Verify agents exchange information securely and with proper authorization via the A2A protocol.
- Ensure agents successfully interact with external data sources using MCP-enabled tools.
- Assess if the core intent of the policy change is correctly identified and summarized by GPT-5.
Metrics
- Impact Assessment Granularity · Target 0.85
- Report Actionability · Target 4
- Adaptive Reasoning Budget Utilization · Target 2
Workflow tasks
- Task 1
Input: {"news_articles": "list_of_strings", "official_releases": "list_of_strings"}
Output: {"policy_identified": "boolean", "policy_summary": "string", "core_intent_interpretation": "string"}
Execution harness: Python
- Task 2
Input: {"policy_summary": "string", "stakeholder_group": "string", "core_intent": "string"}
Output: {"impact_level": "string (low|medium|high)", "impact_details": "string", "recommended_actions": "list_of_strings"}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.