Back to challenges
Challenge

Autonomous E-Commerce Fraud Triage Agent with AutoGen & Galileo

Consumer AI agents executing automated online checkouts create novel automated payment fraud and session hijacking vectors. Build a multi-agent defensive triage network using AutoGen and Galileo to detect anomalous purchasing workflows, evaluate risk metrics, and trigger step-up authentication with 95% precision on synthetic agent interaction traces.

CybersecurityHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Design a multi-agent fraud monitoring network using AutoGen while monitoring agent drift and hallucination risks with Galileo.

Delivery guide

How work is evaluated

Evaluation

Evaluates precision of multi-agent fraud detection and quality of risk explanations logged in Galileo.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

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

schema_conformance

Verifies that all output fields match the triage contract schema.

Binary check

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

Dimension 2

fraud_triage_precision

Precision score of true positive fraud calls vs total flagged calls. • target: 0.95 • range: 0-1

Binary check

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

Learning goals

What you should walk away with

  • Construct a multi-agent AutoGen conversation topology with specialized fraud inspection agents

  • Integrate Galileo evaluation metrics to monitor agent decision quality and hallucinations

  • Analyze automated consumer agent clickstream and checkout behaviors for anomaly markers

  • Implement step-up biometric authentication triggering rules when fraud confidence exceeds threshold

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic dataset of 1,000 e-commerce checkout session logs generated by legitimate and compromised consumer AI shopping agents.

How this agent runs

Evaluates precision of multi-agent fraud detection and quality of risk explanations logged in Galileo.

Challenge input

JSON array of session telemetry containing API request patterns, purchase velocity, and item values

AutoGen

Microsoft framework for multi-agent conversational workflows.

Galileo

Evaluation platform for monitoring agent drift and hallucination metrics.

Evaluated output

JSON triage outcome specifying risk level, action (ALLOW, STEP_UP, BLOCK), and reasoning

Checks for
  • Verifies that all output fields match the triage contract schema.
  • Precision score of true positive fraud calls vs total flagged calls. • target: 0.95 • range: 0-1
Proof of success
  • Benchmark: AgentFraud-Guard-v1
  • Fraud Triage Precision target: 0.95
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • AutoGen · Required
  • Microsoft AutoGen · Optional
Observation
  • Galileo · Optional
Reward / Eval
  • Galileo · Optional

Evaluation contract

  • schema_conformance · Weight 1
  • fraud_triage_precision · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about Autonomous E-Commerce Fraud Triage Agent with AutoGen & Galileo