Build a Cross-Border Trade & Customs Compliance Agent with Vercel AI SDK
Logistics operators face expensive customs delays when cross-border shipping manifests contain misclassified Harmonized System (HS) code descriptions. Build a streaming TypeScript agent using Vercel AI SDK to analyze bill-of-lading documents, validate HS tariff codes, and flag import compliance issues in real time.
What you are building
The core problem, expected build, and operating context for this challenge.
Create a full-stack TypeScript agent with Vercel AI SDK to automate customs manifest checks and cross-border trade document validation.
How work is evaluated
Evaluates Vercel AI SDK agent ability to stream line-item validation and detect invalid HS codes in shipping manifests.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
hs_code_verification_pass
Ensures correct HS code match clears the manifest without error flags
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
classification_accuracy
Accuracy of trade restriction detection across test manifest batch • target: 0.95 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Build dynamic full-stack TypeScript agents using Vercel AI SDK (`ai` package)
Implement streaming UI tool call feedback loops with `streamText` and `generateObject`
Parse shipping manifest line items and query ASEAN Single Window HS code tables
Generate structured trade compliance clearance certificates with error logs
Reference links and supporting material
Synthetic ASEAN bills-of-lading and Harmonized System tariff reference tables.
Challenge setup
Evaluates Vercel AI SDK agent ability to stream line-item validation and detect invalid HS codes in shipping manifests.
Challenge input
JSON object with manifest_id, origin_country, destination_country, items array
AI SDK
Offers typed structured object generation and streaming UI tools
Evaluated output
JSON object with status, clearable, flagged_items array containing hs_code_match
- Ensures correct HS code match clears the manifest without error flags
- Accuracy of trade restriction detection across test manifest batch • target: 0.95 • range: 0-1
- Benchmark: LogiTrade-Bench
- Classification Accuracy target: 0.95
- 1 public reference case
- 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
- Vercel AI SDK · Required
- Zed · Optional
- Evidently AI · Optional
- Vercel AI SDK · Required
Evaluation contract
- hs_code_verification_pass · Weight 1
- classification_accuracy · Weight 1
Metrics
- classification_accuracy · Target 0.95
Benchmarks
- LogiTrade-Bench
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object with manifest_id, origin_country, destination_country, items array
Output: JSON object with status, clearable, flagged_items array containing hs_code_match
Execution harness: JavaScript
Evaluation tools
- AI SDK · Streaming Agent SDK
Recipe freeze marker
No recipe freeze time is recorded.