Skip to main content
Back to challenges
Challenge

Cyber Exposure Triage Agent with OpenAI Agents SDK

Financial firm data breaches require instant triage of cyber policy exposure profiles. You will build an autonomous cyber incident response agent using OpenAI Agents SDK to extract breach indicators, query policy database tools, and compute business interruption potential. The agent must process incident payloads in under 5 seconds with >0.90 classification F1 score.

Machine LearningHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Build an automated cyber security breach triage agent using OpenAI Agents SDK and tool-calling capabilities.

Delivery guide

How work is evaluated

Evaluation

Evaluates threat classification accuracy and speed on simulated breach notification logs.

Datasets

Challenge datasets

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

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

valid_severity

Validates severity level is one of allowable enum values.

Binary check

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

Dimension 2

classification_f1

F1 score on severity classification • target: 0.9 • 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

  • Implement OpenAI Agents SDK runner functions with function calling schema validation.

  • Parse raw incident notification logs to extract compromised PII volume and ransomware indicators.

  • Determine dynamic retention limits and aggregate loss potential.

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of 300 cyber security breach notices with annotated ground-truth reserve recommendations and risk tiers.

Challenge setup

Evaluates threat classification accuracy and speed on simulated breach notification logs.

Challenge input

JSON object with incident_id, breach_summary, records_exfiltrated, encrypted_servers

OpenAI Agents SDK

Assigned framework for autonomous tool-calling agent implementation.

Evaluated output

JSON object with severity_level (CRITICAL|HIGH|MEDIUM|LOW), estimated_loss, recommended_reserve

Evaluation checks
  • Validates severity level is one of allowable enum values.
  • F1 score on severity classification • target: 0.9 • range: 0-1
Configured evaluation evidence
  • Benchmark: Cyber Risk Triage Benchmark
  • Classification F1 target: 0.9
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

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

Action Space
  • OpenAI · Required
  • Akash Network · Optional
  • Darts (Time Series) · Optional

Evaluation contract

  • valid_severity · Weight 1
  • classification_f1 · Weight 1

Metrics

  • classification_f1 · Target 0.9

Benchmarks

  • Cyber Risk Triage Benchmark
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with incident_id, breach_summary, records_exfiltrated, encrypted_servers

    Output: JSON object with severity_level (CRITICAL|HIGH|MEDIUM|LOW), estimated_loss, recommended_reserve

    Execution harness: Python

Evaluation tools

  • OpenAI Agents SDK · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Cyber Exposure Triage Agent with OpenAI Agents SDK