Skip to main content
Loading navigation
Back to challenges

Credit Card Default Risk & Intervention Agent with OpenAI Agents SDK

UK credit card defaults have surged to a 17-year high, requiring financial institutions to proactively identify distressed cardholders before charge-off occurs. Build an early intervention risk management agent using OpenAI Agents SDK to analyze consumer debt accumulation signals and generate customized repayment assistance plans, reducing default rates by 18%.

CybersecurityHosted by Vera
Sign in free to start

What you are building

Implement an autonomous credit monitoring agent using OpenAI Agents SDK with function calling to assess consumer balance trends and trigger timely debt restructuring plans.

How work is evaluated

Evaluation

Evaluates accuracy of default prediction risk tiering and compliance of generated intervention plans.

Datasets

Public datasets, plus the private datasets you have access to.

How submissions are scored

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

FCA Compliance Check

Ensure generated plans do not propose punitive interest rates above legal limits.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

default_prediction_accuracy

Accuracy in categorizing high-risk default borrowers • target: 0.92 • range: 0.85-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

  • Implement OpenAI Agents SDK tools for real-time credit utilization analysis

  • Design personalized debt forbearance strategy recommendations

  • Establish automated customer outreach triggers based on credit score trend vectors

Resources

Dataset notes

Synthetic panel dataset of 10,000 UK consumer credit card account histories tracking 24 months of payment behavior.

Challenge setup

Evaluates accuracy of default prediction risk tiering and compliance of generated intervention plans.

Challenge input

JSON object with monthly balances, payment history, and credit utilization

OpenAI Agents SDK

Mandated agentic framework for OpenAI models.

Evaluated output

JSON output with assigned risk tier (LOW/MEDIUM/HIGH/CRITICAL) and proposed plan

Evaluation checks
  • Ensure generated plans do not propose punitive interest rates above legal limits.
  • Accuracy in categorizing high-risk default borrowers • target: 0.92 • range: 0.85-1
Configured evaluation evidence
  • Benchmark: CreditRisk-Bench
  • Default Prediction Accuracy target: 0.92
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • OpenAI · Required
  • Azure OpenAI Service · Optional
  • OpenAI o3 · Optional

Evaluation contract

  • FCA Compliance Check · Weight 1
  • default_prediction_accuracy · Weight 1

Metrics

  • default_prediction_accuracy · Target 0.92

Benchmarks

  • CreditRisk-Bench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with monthly balances, payment history, and credit utilization

    Output: JSON output with assigned risk tier (LOW/MEDIUM/HIGH/CRITICAL) and proposed plan

    Execution harness: Python

Evaluation tools

  • OpenAI Agents SDK · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Credit Card Default Risk & Intervention Agent with OpenAI Agents SDK