Build GPT-5 & LangGraph Agents for Cross-Border Treasury Automation
The challenge involves developing a sophisticated multi-agent system to automate and optimize cross-border treasury management. Participants will design and implement a LangGraph-based workflow where specialized agents, powered by GPT-5 Pro, collaborate to monitor cash flows, detect anomalies, ensure compliance with international financial regulations, and generate insightful reports. The system will integrate with simulated external financial systems using MCP servers for secure and reliable data exchange. This solution will emphasize extended thinking capabilities for complex financial scenarios, hybrid reasoning (combining instant alerts with deep analysis), and graph-based state management to track and manage the lifecycle of financial transactions and compliance checks. The goal is to build a robust, scalable, and auditable system capable of handling high volumes of cross-border transactions while maintaining accuracy and regulatory adherence.
What you are building
The core problem, expected build, and operating context for this challenge.
Design and implement a LangGraph-based multi-agent system with GPT-5 Pro agents for automated cross-border treasury management, leveraging MCP for enterprise integration and extended thinking for financial analysis.
How work is evaluated
The solution will be evaluated based on the functional correctness of the LangGraph workflow, the effectiveness of agent collaboration, the accuracy of financial analysis and compliance reporting, and the successful integration with simulated MCP tools.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Ensure auditability and explainability in an automated financial system.
Design and implement graph-based agent workflows using LangGraph for complex financial processes.
Integrate LLM-powered agents (GPT-5 Pro) with external financial tools via a simulated MCP.
Orchestrate multi-agent collaboration for tasks like anomaly detection and report generation.
Develop extended thinking strategies for agents to perform in-depth financial analysis and compliance checks.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The solution will be evaluated based on the functional correctness of the LangGraph workflow, the effectiveness of agent collaboration, the accuracy of financial analysis and compliance reporting, and the successful i...
Challenge input
{ "transaction_id": "string", "amount": float, "currency": "string", "sender_country": "string", "receiver_country": "string", "purpose": "string",...
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{ "transaction_id": "string", "compliance_status": "PASS|FAIL", "compliance_report": "string", "flagged_issues": ["string"], "agent_trace": ["strin...
- Verifies if the `compliance_status` (PASS/FAIL) matches the expected outcome based on input rules.
- Checks if `is_anomaly` correctly identifies the anomaly based on the provided transaction history and curre...
- Confirms that the simulated MCP tool calls were correctly made and returned expected (simulated) data.
- ComplianceReportCompleteness 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
- Verifies if the `compliance_status` (PASS/FAIL) matches the expected outcome based on input rules.
- Checks if `is_anomaly` correctly identifies the anomaly based on the provided transaction history and current transaction.
- Confirms that the simulated MCP tool calls were correctly made and returned expected (simulated) data.
Metrics
- ComplianceReportCompleteness · Target 0.9
- ReasoningDepth · Target 0.85
- AgentTraceClarity · Target 0.9
Workflow tasks
- Task 1
Input: { "transaction_id": "string", "amount": float, "currency": "string", "sender_country": "string", "receiver_country": "string", "purpose": "string", "compliance_rules": ["string"] }
Output: { "transaction_id": "string", "compliance_status": "PASS|FAIL", "compliance_report": "string", "flagged_issues": ["string"], "agent_trace": ["string"] }
Execution harness: Python
- Task 2
Input: { "transaction_history": [{"timestamp": "datetime", "amount": float, "currency": "string", "sender_country": "string", "receiver_country": "string", "purpose": "string"}], "current_transaction": {"timestamp": "datetime", "amount": float, "currency": "string", "sender_country": "string", "receiver_country": "string", "purpose": "string"} }
Output: { "is_anomaly": boolean, "anomaly_reason": "string|null", "confidence_score": float, "agent_trace": ["string"] }
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.