ClawHub
Back to skill

Security audit

lv-guide-display

Security checks across malware telemetry and agentic risk

Overview

This is a messy calendar-helper skill with an overbroad trigger, but the artifact is only markdown and does not install code, request credentials, access private calendars, or run commands.

Install only if you are comfortable with a noisy calendar helper. The publisher should clean up the repeated metadata, narrow the trigger, and clearly disclose whether schedule management is only conversational or writes anywhere persistent.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The manifest advertises a narrow read-only calendar lookup skill, but the body documents broader capabilities including schedule creation and management. This mismatch can mislead reviewers, users, and policy gates about what the skill can do, increasing the risk of unreviewed data-handling behavior or unintended privilege use.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The declared trigger says the skill should activate only for mentions of today's calendar, but the documented intents cover tomorrow, holidays, countdowns, and schedule management. This creates scope drift and can cause the skill to be invoked in contexts users did not intend, potentially exposing calendar or schedule functionality during ordinary conversation.

Vague Triggers

Medium
Confidence
89% confidence
Finding
A broad trigger like mentioning today's calendar is ambiguous and likely to overlap with normal conversation, causing accidental invocation. In an assistant environment, unintended activation can route user queries to this skill without clear consent or expectation, especially if the skill has broader behavior than advertised.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The repeated activation text reinforces an imprecise invocation condition but never defines limits, exclusions, or disambiguation rules. That ambiguity increases the chance of false activations and, combined with the documented expanded feature set, broadens the practical attack surface of the skill.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.