Claude Agents SDK & Sourcegraph Amp Luxury Merchant Verification
Premium e-commerce platforms struggle with digital public infrastructure fraud and counter-fake verification on open networks. You will develop an autonomous inspection agent that cross-references brand registries, tax filings, and product authenticity documentation to automate luxury merchant trust scoring.
What you are building
The core problem, expected build, and operating context for this challenge.
Construct a Claude Agents SDK workflow integrated with Sourcegraph Amp repository inspection tools to dynamically audit merchant compliance and product authenticity metrics.
How work is evaluated
Evaluates seller authenticity audit precision across simulated business registries.
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.
Invalid GSTIN Rejection
Must reject sellers with invalid or unverified GSTIN formats
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Verification Precision
Accuracy in detecting fraudulent trademark claims • target: 0.96 • range: 0.88-1
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 custom tool calls in Claude Agents SDK for GSTIN and trademark registry auditing
Integrate Sourcegraph Amp workspace automation to inspect code repositories and API integration points
Construct a dynamic risk scoring engine for seller onboarding on open e-commerce rails
Enforce compliance checks for high-value fashion and luxury catalog listings
Reference links and supporting material
Synthetic merchant filings containing GSTIN numbers, brand trademark registry entries, and catalog listings.
Challenge setup
Evaluates seller authenticity audit precision across simulated business registries.
Challenge input
JSON with seller business name, GSTIN, trademark registration number, and repo URL
Claude Agents SDK
Provides reasoning and structured tool interaction.
Sourcegraph Amp
Inspects merchant integration codebase for security compliance.
Evaluated output
JSON with trust_score (0-100), verification_status, and risk_flags array
- Must reject sellers with invalid or unverified GSTIN formats
- Accuracy in detecting fraudulent trademark claims • target: 0.96 • range: 0.88-1
- Benchmark: MerchantRisk-Bench
- Verification Precision target: 0.96
- 1 public reference case
- 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
- Sourcegraph Amp · Required
- Libretto · Optional
- Retell AI · Optional
- Sourcegraph Amp · Required
Evaluation contract
- Invalid GSTIN Rejection · Weight 1
- Verification Precision · Weight 1
Metrics
- Verification Precision · Target 0.96
Benchmarks
- MerchantRisk-Bench
- security
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with seller business name, GSTIN, trademark registration number, and repo URL
Output: JSON with trust_score (0-100), verification_status, and risk_flags array
Execution harness: Python
Evaluation tools
- Claude Agents SDK · Agent Framework
- Sourcegraph Amp · Repository Audit
Recipe freeze marker
No recipe freeze time is recorded.