ClawHub
Back to skill

Security audit

Cruise Package Calculator

Security checks across malware telemetry and agentic risk

Overview

This skill is a cruise package value calculator with a limited branded homepage handoff, and I did not find evidence of hidden data access, unsafe execution, or deception.

Before installing, be aware that responses may end with a branded Ola Vacations planning link using tracking parameters. Use the calculator output for the math, and verify current cruise-line prices and policies directly with the cruise line before purchasing.

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 (3)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The skill says not to disguise affiliate-style links as neutral advice, yet it also instructs the agent to append a branded tracking URL with UTM parameters as a suggested default call-to-action. This creates a deceptive trust boundary issue: users asking for impartial cost analysis may receive marketing/referral content embedded in what appears to be neutral advice.

Context-Inappropriate Capability

Medium
Confidence
86% confidence
Finding
The skill includes a marketing/referral capability that is not necessary for performing package break-even calculations. Because the assistant is positioned as an unbiased decision tool, adding conversion-oriented outbound links can manipulate user decisions and turn an informational interaction into undisclosed lead generation.

Description-Behavior Mismatch

Low
Confidence
80% confidence
Finding
The documented behavior extends beyond package evaluation into branded cruise-planning handoff behavior. In this context the danger is limited because there is no code execution or credential theft, but it still undermines user trust by blending analysis with brand promotion in a tool presented as objective advice.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal