Tactical Intelligence Agent System
Inspired by the use of AI drones in real-world scenarios, this challenge focuses on building a sophisticated multi-agent system for real-time tactical intelligence analysis. Participants will design and implement a distributed network of agents capable of ingesting diverse data streams (e.g., simulated sensor feeds, news articles, public reports) and performing rapid, ethical analysis to provide strategic recommendations. The system must leverage the MCP for secure and standardized tool integration, allowing agents to interact with external data sources and decision-making tools in a controlled environment. This system will utilize a graph-based workflow orchestrator to manage complex reasoning paths and agent-to-agent communication via the A2A protocol. Agents will employ extended thinking techniques with adaptive reasoning budgets to prioritize critical analysis under time constraints, ensuring high-fidelity outputs while maintaining computational efficiency. The ultimate goal is to demonstrate an intelligent, autonomous system that can process ambiguous information, identify potential threats or opportunities, and offer actionable insights, mimicking the operational demands of advanced AI systems in sensitive applications.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a graph-based multi-agent system using LangGraph with GPT-5 and MCP for autonomous tactical intelligence analysis.
How work is evaluated
The evaluation will assess the system's ability to accurately process information, generate coherent tactical recommendations, and demonstrate proper MCP tool integration and A2A communication.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Deploy a resilient agent system utilizing Factory AI principles for manageability.
Establish robust A2A protocol communication for inter-agent collaboration.
Develop agents employing extended thinking with adaptive reasoning budgets.
Integrate MCP for secure and standardized tool access to external data sources.
Design and implement a graph-based multi-agent system for tactical intelligence.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation will assess the system's ability to accurately process information, generate coherent tactical recommendations, and demonstrate proper MCP tool integration and A2A communication.
Challenge input
{'scenario_description': 'string', 'data_feeds': [{'type': 'string', 'content': 'string'}]}
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{'summary': 'string', 'recommendations': ['string'], 'confidence_score': 'float'}
- Ensures recommendations are present and in the correct array format.
- Verifies that at least one MCP-enabled tool was successfully invoked during scenario analysis.
- Checks logs for evidence of successful A2A protocol messages between designated agents.
- RecommendationAccuracy 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
- Ensures recommendations are present and in the correct array format.
- Verifies that at least one MCP-enabled tool was successfully invoked during scenario analysis.
- Checks logs for evidence of successful A2A protocol messages between designated agents.
Metrics
- RecommendationAccuracy · Target 0.85
- AdaptiveBudgetEfficiency · Target 0.75
Workflow tasks
- Task 1
Input: {'scenario_description': 'string', 'data_feeds': [{'type': 'string', 'content': 'string'}]}
Output: {'summary': 'string', 'recommendations': ['string'], 'confidence_score': 'float'}
Execution harness: Python
- Task 2
Input: {'data_point': {'key': 'value'}, 'expected_tool_call': 'string'}
Output: {'tool_called': 'boolean', 'tool_output': 'json'}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.