Skip to main content
Back to challenges
Challenge

TypeScript AI SDK Cyber Exposure Risk Model for Acquisitions

Cyber insurance underwriters need fast risk scoring engines to evaluate enterprise security postures prior to portfolio acquisition. Build a TypeScript Vercel AI SDK workflow to assess cyber vulnerability scan reports and predict expected annual loss (EAL).

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.

Construct an AI SDK stream-enabled pipeline in TypeScript to analyze network telemetry and quantify cyber M&A exposure.

Delivery guide

How work is evaluated

Evaluation

Evaluates precision of extracted vulnerability severity ratings and expected financial loss calculations.

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

zod_schema_match

Ensures output validates against Zod contract schema

Binary check

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

Dimension 2

eal_accuracy

Accuracy of loss estimate compared to actuarial model • target: 0.88 • 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

  • Use Vercel AI SDK streamText and generateObject functions in TypeScript

  • Parse external attack surface vulnerability scans (CVE lists, open ports)

  • Calculate Cyber Expected Annual Loss (EAL) using structured LLM output schemas

Resources and assets

Reference links and supporting material

Dataset notes

Cyber vulnerability scan dataset of 300 target enterprise networks with ground truth loss histories.

Challenge setup

Evaluates precision of extracted vulnerability severity ratings and expected financial loss calculations.

Challenge input

JSON with cve_list and asset_count

AI SDK

Vercel's framework for streamable, type-safe AI workflows

Evaluated output

JSON containing eal_usd, risk_grade, and key_vulnerabilities

Evaluation checks
  • Ensures output validates against Zod contract schema
  • Accuracy of loss estimate compared to actuarial model • target: 0.88 • range: 0-1
Configured evaluation evidence
  • Benchmark: Cyber-Risk-Bench
  • Eal Accuracy target: 0.88
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript 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
  • TensorRT-LLM · Required
  • Writer · Optional
Policy Serving
  • Qwen 3 · Optional

Evaluation contract

  • zod_schema_match · Weight 1
  • eal_accuracy · Weight 1

Metrics

  • eal_accuracy · Target 0.88

Benchmarks

  • Cyber-Risk-Bench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with cve_list and asset_count

    Output: JSON containing eal_usd, risk_grade, and key_vulnerabilities

    Execution harness: JavaScript

Evaluation tools

  • AI SDK · Agent TypeScript Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about TypeScript AI SDK Cyber Exposure Risk Model for Acquisitions