Back to challenges
Challenge

Fuel Terminal Logistics & Guardrail Copilot with Vercel AI SDK

Managing inventory dispatch for Libya's new fuel terminal contract requires AI assistance that enforces safety protocols and prevents dangerous overfill commands. Build a Web application using Vercel AI SDK and Guardrails AI to validate model outputs before issuing terminal valve control instructions. Achieve zero safety policy violations across test runs.

Business OperationsHosted 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 secure fuel terminal dispatch web API using Vercel AI SDK and Guardrails AI validation.

Delivery guide

How work is evaluated

Evaluation

Evaluates safety compliance and parameter validation for fuel terminal dispatch commands generated via Vercel AI SDK.

Datasets

Shared data for this challenge

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

Loading datasets...
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

Zero Overfill Violation

Fails if approved volume + current volume > max capacity.

Binary check

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

Dimension 2

Guardrail Compliance Rate

Percentage of invalid requests correctly caught and blocked by Guardrails AI. • 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

What you should walk away with

  • Utilize Vercel AI SDK streaming text and tool calls in TypeScript

  • Apply Guardrails AI rules to block unsafe pressure or volume recommendations

  • Implement full exception logging for rejected dangerous actions

  • Deliver automated UI status updates to fuel depot supervisors

Resources and assets

Reference links and supporting material

Dataset notes

Fuel storage tank operational datasets and safety limit specifications for Libya fuel terminal.

How this agent runs

Evaluates safety compliance and parameter validation for fuel terminal dispatch commands generated via Vercel AI SDK.

Challenge input

JSON object with tank_id, target_volume_liters, max_capacity_liters, current_volume_liters

AI SDK

Required AI web integration SDK.

Guardrails AI

Prevents catastrophic terminal valve overspills.

Evaluated output

JSON object with status (APPROVED/REJECTED), safe_volume_liters, message

Checks for
  • Fails if approved volume + current volume > max capacity.
  • Percentage of invalid requests correctly caught and blocked by Guardrails AI. • target: 1 • range: 0-1
Proof of success
  • Benchmark: GuardrailBench
  • Guardrail Compliance Rate target: 100%
  • 1 public reference case
Runtime evidence
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • RAI · Required
  • Vercel AI SDK · Optional
Reward / Eval
  • Guardrails AI · Optional
Safety / Guardrails
  • Guardrails AI · Optional
Orchestration
  • Vercel AI SDK · Optional

Evaluation contract

  • Zero Overfill Violation · Weight 1
  • Guardrail Compliance Rate · Weight 1

Recipe state

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

Frequently Asked Questions about Fuel Terminal Logistics & Guardrail Copilot with Vercel AI SDK