ClawHub
Back to skill

Security audit

信用卡推荐

Security checks across malware telemetry and agentic risk

Overview

This skill coherently recommends credit cards from a public bank data feed, but its application links include disclosed channel and platform parameters users should understand before clicking.

Install only if you are comfortable with a credit-card recommendation skill fetching card data from CITIC and showing application links that include channel/referral-style sid parameters and, on recognized clients, a coarse platform tag. Review card fees and terms yourself before applying.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill inspects host environment variables to detect which client platform it is running under, then uses that information to influence outbound application links. For a credit-card recommendation skill, this is unnecessary for core filtering/recommendation behavior and creates an avoidable channel for collecting execution-context metadata that can be used for attribution or tracking.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The code rewrites application URLs by appending sid and platform parameters before presenting them to users. This exceeds a plain recommendation role and introduces tracking/attribution behavior that can leak execution context to the remote bank endpoint and make user clicks traceable to a specific host platform.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.