ClawHub
Back to skill

Security audit

购物比价助手

Security checks across malware telemetry and agentic risk

Overview

The skill is a shopping price comparison tool, but it overstates platform coverage and sends searches through hard-coded Tencent cloud proxy endpoints using a bundled token.

Review this before installing if accurate Pinduoduo coverage or query privacy matters to you. It appears limited to JD/Taobao comparison and does not buy anything, but your search terms are sent to cloud proxy services and the package includes a shared proxy token.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
82% confidence
Finding
The skill declares environment variables for a proxy service and implicitly relies on network access, but no explicit permissions or trust boundaries are documented. This creates a transparency and review gap: users and reviewers cannot clearly assess what external systems receive queries or how secrets such as proxy tokens are handled.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding
The skill advertises full three-platform price comparison including Pinduoduo, but the analyzed behavior reportedly does not support Pinduoduo and instead returns a non-support message. This mismatch is dangerous because it misleads users about coverage and may cause incorrect purchasing decisions, while also undermining trust in what data is actually fetched and compared.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
User search queries are transmitted to external Tencent SCF proxy endpoints, but the skill gives no user-facing disclosure or consent mechanism. In a shopping assistant context, queries can reveal purchasing intent, brand preferences, or other sensitive behavioral data, and the hidden proxy layer increases privacy risk because users may assume processing is local or direct.

Missing User Warnings

Low
Confidence
97% confidence
Finding
The code embeds a default proxy token ("tp_8k2mX9vQ4z") as a fallback when the environment variable is absent. Hard-coded credentials are dangerous because anyone with code access can reuse the token against the proxy endpoints, potentially causing unauthorized use, quota exhaustion, billing abuse, or access to backend services tied to that token.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.