Skip to main content
Back to challenges
Challenge

MCP-Enabled AI for Luxury Authenticity on TikTok Shop

The luxury retail market is expanding rapidly on platforms like TikTok Shop, but with this growth comes the critical challenge of authenticating high-value items, many sold by secondhand resellers. This challenge focuses on building a cutting-edge multi-modal generative AI system that leverages advanced agent frameworks to verify the authenticity of luxury goods. Developers will integrate real-time image analysis, historical data lookup, and supply chain provenance to ensure every item meets stringent authenticity standards. Participants will design a sophisticated LangGraph-based workflow, orchestrating multiple specialized agents to perform detailed checks, cross-reference databases, and identify anomalies. The system will utilize the multi-modal capabilities of Gemini 3 to process visual cues, textual descriptions, and potentially even audio or video input for a comprehensive assessment. Crucially, the solution will incorporate the MCP for seamless tool integration with external enterprise systems and blockchain explorers, ensuring agents can access and process vast amounts of structured and unstructured data for definitive authenticity decisions. This challenge pushes the boundaries of AI in e-commerce, creating a verifiable and trustworthy luxury marketplace.

Special Purpose AgentsHosted 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 graph-based multi-modal agent system using LangGraph with Gemini 3 for luxury product authenticity verification, leveraging MCP tool integration.

Delivery guide

How work is evaluated

Evaluation

The evaluation will assess the authenticity system's accuracy, robustness, and ability to generate comprehensive verification reports. It will simulate various luxury product verification scenarios, including both genuine and counterfeit items.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Learning goals

What you should walk away with

  • Utilize advanced RAG techniques with LlamaIndex for context-rich authenticity analysis.

  • Build a robust, stateful workflow using LangGraph for complex, multi-step authentication processes.

  • Integrate MCP-enabled tools for data retrieval from diverse external systems like blockchain and product databases.

  • Develop a system that provides transparent and auditable verification reports.

  • Design and implement a multi-modal agent system for luxury goods authenticity verification.

Resources and assets

Reference links and supporting material

Dataset notes

Sample data for 2 tasks

Challenge setup

The evaluation will assess the authenticity system's accuracy, robustness, and ability to generate comprehensive verification reports. It will simulate various luxury product verification scenarios, including both gen...

Challenge input

{ 'product_name': 'string', 'description': 'string', 'image_urls': ['string'], 'metadata': { 'serial_number': 'string', 'seller_id': 'string' } }

Agent execution

No agent tool is listed for this challenge.

Evaluated output

{ 'is_authentic': 'boolean', 'confidence_score': 'float', 'evidence_report': ['string'], 'issues_detected': ['string'], 'trace_data': { 'blockchain...

Evaluation checks
  • System correctly identifies authenticity for a diverse set of items (70% accuracy for pass).
  • Evidence report provides at least 3 distinct, relevant points for each verification.
  • Demonstrates successful data retrieval using MCP-enabled tools from simulated external services.
Configured evaluation evidence
  • OverallAccuracy target: 0.85
Configured runtime
  • Python execution harness

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

No tool records are attached.

Evaluation contract

  • System correctly identifies authenticity for a diverse set of items (70% accuracy for pass).
  • Evidence report provides at least 3 distinct, relevant points for each verification.
  • Demonstrates successful data retrieval using MCP-enabled tools from simulated external services.

Metrics

  • OverallAccuracy · Target 0.85
  • ReportDetailScore · Target 4

Workflow tasks

  1. Task 1

    Input: { 'product_name': 'string', 'description': 'string', 'image_urls': ['string'], 'metadata': { 'serial_number': 'string', 'seller_id': 'string' } }

    Output: { 'is_authentic': 'boolean', 'confidence_score': 'float', 'evidence_report': ['string'], 'issues_detected': ['string'], 'trace_data': { 'blockchain_tx': 'string', 'supplier_info': 'string' } }

    Execution harness: Python

  2. Task 2

    Input: { 'product_name': 'string', 'description': 'string', 'image_urls': ['string'], 'metadata': { 'serial_number': 'string', 'seller_id': 'string' } }

    Output: { 'is_authentic': 'boolean', 'confidence_score': 'float', 'evidence_report': ['string'], 'issues_detected': ['string'] }

    Execution harness: Python

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about MCP-Enabled AI for Luxury Authenticity on TikTok Shop