MCP-Enabled Federal Data Agent
Inspired by the 'Genesis Mission' to boost AI innovation using federal datasets, this challenge tasks you with building a sophisticated agentic system. You will design and implement a graph-based multi-agent system using LangGraph to autonomously access, analyze, and synthesize scientific information from simulated federal datasets. The system will leverage GPT-5 for advanced reasoning and LlamaIndex for robust Retrieval-Augmented Generation (RAG) capabilities, ensuring accurate and contextual information retrieval. A core component will be the integration of MCP-enabled tools for seamless and standardized interaction with various simulated enterprise APIs representing federal data sources, demonstrating how agents can securely and efficiently work with structured and unstructured governmental data for novel scientific insights.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a graph-based multi-agent system using LangGraph, GPT-5, and LlamaIndex for autonomous scientific research on federal datasets, integrating with MCP-enabled tools.
How work is evaluated
The evaluation will assess the agent system's ability to accurately extract information, generate coherent scientific hypotheses, and demonstrate robust MCP tool integration.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Design and implement graph-based agent workflows using LangGraph for complex scientific inquiries.
Develop and utilize MCP-enabled tool integration for standardized access to external enterprise data sources.
Integrate LlamaIndex with vector databases to perform advanced RAG over diverse federal datasets.
Leverage GPT-5's extended thinking capabilities to generate novel scientific hypotheses and insights.
Reference links and supporting material
Sample data for 3 tasks
Challenge setup
The evaluation will assess the agent system's ability to accurately extract information, generate coherent scientific hypotheses, and demonstrate robust MCP tool integration.
Challenge input
{'dataset_id': 'unique_id', 'query': 'natural language question'}
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{'extracted_data': {'key': 'value'}, 'summary': 'string', 'confidence': float}
- Ensure the LangGraph workflow executes without unhandled errors.
- Verify successful invocation of at least one MCP-enabled tool.
- Data Extraction F1 Score 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
- Ensure the LangGraph workflow executes without unhandled errors.
- Verify successful invocation of at least one MCP-enabled tool.
Metrics
- Data Extraction F1-Score · Target 0.85
- Hypothesis Plausibility · Target 0.7
- MCP Integration Latency · Target 100
Workflow tasks
- Task 1
Input: {'dataset_id': 'unique_id', 'query': 'natural language question'}
Output: {'extracted_data': {'key': 'value'}, 'summary': 'string', 'confidence': float}
Execution harness: Python
- Task 2
Input: {'facts': ['fact1', 'fact2', ...], 'domain': 'string'}
Output: {'hypothesis': 'string', 'justification': 'string', 'novelty_score': float}
Execution harness: Python
- Task 3
Input: {'tool_name': 'string', 'parameters': {'param1': 'value1'}}
Output: {'status': 'success'|'failure', 'response_data': 'json_or_error_message'}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.