ClawHub
Back to skill

Security audit

Travel Assistant Tool Free

Security checks across malware telemetry and agentic risk

Overview

This travel-planning skill is mostly ordinary, but it asks for shell and file-writing authority while its instructions are inconsistent about export behavior and when the skill should activate.

Install only if you are comfortable with a Markdown travel assistant that may use shell/network access for weather checks. Treat file export as unsupported unless the publisher clarifies the contradiction, and avoid granting write or exec permissions where your agent allows narrower tool scoping.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The skill declares `write` and `exec` tools and documents export/save/convert operations, while other sections state the free edition cannot export files. This mismatch can mislead users and downstream agents into performing filesystem writes or generating artifacts the user did not expect, expanding the skill’s effective capability beyond its stated trust boundary.

Intent-Code Divergence

Medium
Confidence
86% confidence
Finding
The documentation advertises export/save operations but the FAQ later says the free version only supports in-chat viewing and not file export. Such contradictory instructions create unsafe ambiguity for autonomous agents, which may choose the more permissive path and write files or transform data unexpectedly.

Vague Triggers

High
Confidence
90% confidence
Finding
The trigger conditions describe translation and localization use cases unrelated to this travel skill, which can cause over-broad or incorrect activation. In an agent environment with `exec` and networked behavior, unintended activation increases the chance the skill runs commands or accesses remote services in contexts where the user did not intend to use this tool.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill documents network queries and command execution, including `curl` and connectivity testing, but does not clearly warn users about command execution, external data transfer, logging, or the trust implications of contacting third-party services. This omission can lead to unconsented outbound requests or shell usage in environments where users expect a purely conversational planning tool.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.