ClawHub
Back to skill

Security audit

跨境电商选品工具

Security checks across malware telemetry and agentic risk

Overview

The code is mostly a coherent e-commerce analysis tool, but its declared purchase/crypto capability is not explained by the artifacts.

Review the requested crypto and purchase capability tags before installing, because they are not justified by the visible code. Use the tool as a local MVP with simulated market data, avoid exposing the Flask debug server publicly, and only configure an OpenAI API key if sending product and market context to OpenAI is acceptable.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The code reads an API key from the environment and sends product_name, keyword_analysis, and competitor_data to OpenAI without any visible consent gate, data-classification check, or scope restriction. In a skill context with no declared metadata or documented external-sharing purpose, this creates a real data exposure risk because potentially sensitive business information can be transmitted to a third-party service.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The service presents itself as scraping real competitor data, but it only fabricates randomized mock products. This is dangerous because downstream users or agents may make business decisions based on false market intelligence, and the misleading docstrings/comments conceal that the outputs are synthetic rather than real.

Natural-Language Policy Violations

Medium
Confidence
92% confidence
Finding
The skill content is entirely in Chinese and presents commands, descriptions, and outputs without any indication of multilingual support or user language selection. This can cause users to misunderstand commands, outputs, pricing, or platform details, which is a real quality and safety issue, though it does not directly enable code execution or privilege abuse.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The UI collects user-entered keywords and product details and transmits them to backend API endpoints, but the page shows no notice that this data will be sent, processed, or potentially stored. In a product research tool, these inputs may contain commercially sensitive market intelligence, so the lack of transparency creates a real privacy and trust issue even though it is not an exploit primitive like XSS or RCE.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.