ClawHub
Back to skill

Security audit

智能灯控(免费版)

Security checks across malware telemetry and agentic risk

Overview

This appears to be a local smart-light control skill, but its unrelated trigger wording and immediate device-control commands make it worth reviewing before use.

Install only if you intend to let an agent run local-network smart-light commands. Before using it, narrow the trigger to smart-light tasks only, confirm the target bulb IP and requested change before execution, and avoid automatic use from unrelated data-analysis or reporting requests.

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
92% confidence
Finding
The skill advertises concrete LAN smart-light control, but the body in this section is largely generic workflow boilerplate using placeholder parameter names and abstract CRUD-style operations instead of precise device-control instructions. This mismatch can cause an agent or user to invoke the skill under false assumptions, increasing the chance of unintended command execution, incorrect automation behavior, or trust in unsupported functionality.

Description-Behavior Mismatch

Medium
Confidence
97% confidence
Finding
The trigger condition says the skill should be used for data analysis, report generation, statistics, and visualization, which is unrelated to smart-light control. Overbroad and incorrect routing criteria can cause an agent to select this skill in the wrong context, potentially leading to unnecessary exec usage or accidental physical device actions when the user's task is unrelated.

Vague Triggers

High
Confidence
98% confidence
Finding
The trigger condition is both overly broad and unrelated to the skill's real purpose, making inappropriate activation likely. In a skill with exec capability and physical device-control semantics, bad routing is more dangerous because it can prompt command execution or state-changing actions in contexts where the user never intended to control hardware.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The documentation instructs immediate device-control actions but does not prominently warn that commands will change the state of a real physical device on the local network. In the context of a hardware-control skill with exec access, missing confirmation and safety messaging increases the risk of unintended activation, disruption, or nuisance actions against devices reachable on the LAN.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.