Skip to main content
Back to challenges
Challenge

Multi-Agent Reinsurance Portfolio Treaty Analysis with CrewAI and GitHub Copilot

Global reinsurers manage massive exposure concentrations across primary carriers and catastrophe treaties. You will create a CrewAI agent team assisted by GitHub Copilot code generation to analyze capital reserves, catastrophe risk limits, and treaty terms across Munich Re and Lloyd's portfolios. The multi-agent crew must correctly identify portfolio treaty overlaps with 90%+ accuracy.

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 a collaborative CrewAI multi-agent workforce leveraging GitHub Copilot tooling to evaluate global reinsurance treaty exposure and portfolio accumulation.

Delivery guide

How work is evaluated

Evaluation

Evaluates multi-agent CrewAI output for reinsurance treaty concentration accuracy and exposure limit compliance.

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

check_exposure_math

Accumulated exposure must equal sum of individual limits.

Binary check

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

Dimension 2

accuracy_score

Accuracy in identifying treaty breach conditions. • 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

  • Configure role-playing agents in CrewAI for Treaty Analysis, Exposure Accumulation, and Capital Allocation.

  • Integrate custom financial aggregation tools assisted by GitHub Copilot boilerplate logic.

  • Execute multi-turn agent delegation to evaluate Lloyd's syndicate catastrophe commitments.

  • Generate actionable treaty optimization reports with auditable capital allocation metrics.

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic dataset of 100 global reinsurance treaties modeling ceding limits, retentions, and hazard triggers across top tier global reinsurers.

Challenge setup

Evaluates multi-agent CrewAI output for reinsurance treaty concentration accuracy and exposure limit compliance.

Challenge input

JSON array of treaties containing reinsurer_name, ceding_carrier, limit_usd, retention_usd, hazard_type

CrewAI

Orchestrates multi-agent role-playing workflows for complex reinsurance portfolio reviews.

GitHub Copilot

Accelerates creation of custom domain tools and mathematical helper utilities.

Evaluated output

JSON object with accumulated_exposure, limit_breaches (ARRAY), and recommended_hedges (ARRAY)

Evaluation checks
  • Accumulated exposure must equal sum of individual limits.
  • Accuracy in identifying treaty breach conditions. • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: ReinsuranceAgentBench-2026
  • Accuracy Score target: 0.95
  • 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
  • GitHub Copilot · Required
  • GitHub · Optional
  • CrewAI · Optional
Observation
  • GitHub Copilot · Required

Evaluation contract

  • check_exposure_math · Weight 1
  • accuracy_score · Weight 1

Metrics

  • accuracy_score · Target 0.95

Benchmarks

  • ReinsuranceAgentBench-2026
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON array of treaties containing reinsurer_name, ceding_carrier, limit_usd, retention_usd, hazard_type

    Output: JSON object with accumulated_exposure, limit_breaches (ARRAY), and recommended_hedges (ARRAY)

    Execution harness: Python

Evaluation tools

  • CrewAI · Multi-Agent Framework
  • GitHub Copilot · AI Code Assistant

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Multi-Agent Reinsurance Portfolio Treaty Analysis with CrewAI and GitHub Copilot