Skip to main content
Loading navigation
Back to challenges

Stateful Energy Export Dispute & Claims Handling Agent using Mastra AI

Short-term export approvals can lead to commercial disputes over laytime delays, crude specification off-specs, and tax demurrage claims. Build a durable, stateful dispute resolution workflow agent using Mastra AI. The agent uses Mastra AI's built-in vector memory, stateful step workflows, and automated system logging to process demurrage claims, cross-check port log telemetry, and determine contractual liability limits.

Business OperationsHosted by Vera
Sign in free to start

What you are building

Construct a stateful, resilient claims and dispute resolution agent using Mastra AI engine.

How work is evaluated

Evaluation

Evaluates demurrage claim calculation accuracy and fault determination.

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

demurrage_math_check

Ensure total demurrage calculation is exact.

Binary check

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

Dimension 2

workflow_execution_success

Success rate of step workflow state completion. • target: 1 • 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

  • Implement stateful workflow steps with Mastra AI in TypeScript.

  • Use Mastra AI vector memory to retrieve precedent demurrage claim rulings.

  • Generate automated liability determination reports.

Resources

Dataset notes

Dataset of 30 shipping demurrage claims and charter party clauses.

Challenge setup

Evaluates demurrage claim calculation accuracy and fault determination.

Challenge input

JSON with allowed_laytime_hours, actual_loading_hours, demurrage_rate_per_day

Mastra AI

TypeScript framework supporting resilient workflows and agent memory.

Evaluated output

JSON with excess_hours, total_demurrage_usd, fault_party

Evaluation checks
  • Ensure total demurrage calculation is exact.
  • Success rate of step workflow state completion. • target: 1 • range: 0-1
Configured evaluation evidence
  • Benchmark: Mastra-Workflow-Bench
  • Workflow Execution Success target: 1
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript 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
  • Aim · Required
  • Mastra AI · Optional
  • Durable · Optional
Orchestration
  • Mastra AI · Optional

Evaluation contract

  • demurrage_math_check · Weight 1
  • workflow_execution_success · Weight 1

Metrics

  • workflow_execution_success · Target 1

Benchmarks

  • Mastra-Workflow-Bench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with allowed_laytime_hours, actual_loading_hours, demurrage_rate_per_day

    Output: JSON with excess_hours, total_demurrage_usd, fault_party

    Execution harness: JavaScript

Evaluation tools

  • Mastra AI · Stateful Orchestration

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Stateful Energy Export Dispute & Claims Handling Agent using Mastra AI