Build a Factual Integrity Agent System
The proliferation of disinformation, particularly the 'LLM grooming' tactics necessitates robust AI systems for factual verification. This challenge tasks you with building an advanced multi-agent system designed to detect, analyze, and counter sophisticated disinformation campaigns. Your system will leverage cutting-edge large language models and a graph-based agent framework to establish a factual integrity pipeline. Agents will employ extended thinking and adaptive reasoning budgets to critically evaluate information, cross-reference with multiple authoritative sources via MCP-enabled tools, and communicate securely to synthesize verified insights. This project emphasizes developing resilient and context-aware agents capable of distinguishing nuanced truth from elaborate falsehoods in real-time, preparing them for the challenges of an information-saturated digital landscape.
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 Claude Opus 4.1 for autonomous factual verification and disinformation detection.
How work is evaluated
Evaluation will focus on the system's ability to accurately detect and explain disinformation, its resilience to subtle manipulations, and the efficiency of its agentic workflow.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Integrate Model Context Protocol (Model Context Protocol) for seamless tool access to authoritative knowledge bases and fact-checking APIs.
Develop agents that employ extended thinking with adaptive reasoning budgets to handle nuanced disinformation.
Design and implement a graph-based multi-agent architecture using LangGraph for complex factual verification workflows.
Orchestrate agent-to-agent communication for collaborative information analysis and synthesis.
Implement advanced RAG techniques to retrieve and evaluate information from diverse, verifiable sources.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
Evaluation will focus on the system's ability to accurately detect and explain disinformation, its resilience to subtle manipulations, and the efficiency of its agentic workflow.
Challenge input
{ "text": "string", "topic": "string" }
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{ "disinformation_detected": "boolean", "identified_claims": [ { "claim": "string", "is_disinformation": "boolean", "evidence": "string", "explanat...
- System correctly identifies major disinformation elements in at least 80% of test cases.
- Generated explanations for disinformation are coherent, well-supported by evidence, and accurate.
- Precision Of Disinformation Detection target: 0.9
- 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
- System correctly identifies major disinformation elements in at least 80% of test cases.
- Generated explanations for disinformation are coherent, well-supported by evidence, and accurate.
Metrics
- Precision of Disinformation Detection · Target 0.9
- Reasoning Efficiency · Target 4
Workflow tasks
- Task 1
Input: { "text": "string", "topic": "string" }
Output: { "disinformation_detected": "boolean", "identified_claims": [ { "claim": "string", "is_disinformation": "boolean", "evidence": "string", "explanation": "string" } ], "confidence": "float" }
Execution harness: Python
- Task 2
Input: { "query": "string", "max_iterations": "int" }
Output: { "final_answer": "string", "reasoning_steps": "int", "tool_calls": ["string"] }
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.