ClawHub
Back to skill

Security audit

投屏控制专业版

Security checks for vulnerabilities and agentic risk

Overview

This skill is mainly a Chromecast control skill, but its trigger text and permissions are too broad for actions that can affect real devices on a local network.

Review this carefully before installing. It should only be used when you intend to let an agent discover and control Chromecast devices on your local network, including group casting and scheduled stop actions. Avoid using it for project management or collaboration requests despite the manifest text, and require explicit confirmation before any cast-to-all, group cast, monitoring, or stop-all operation.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Description-Behavior Mismatch

High
Confidence
97% confidence
Finding
The manifest description claims the skill is for project management, task planning, progress tracking, and team collaboration, while the implementation actually performs Chromecast discovery and device-control operations. This mismatch can cause the agent or user to invoke a network/device-control skill under false pretenses, reducing informed consent and bypassing scrutiny normally applied to operational tooling.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill executes system-level commands via subprocess to scan for devices and control playback, but the manifest frames the tool as a benign productivity/project-management skill. That unjustified capability expansion increases the chance that an agent grants or uses Bash/exec privileges without the user understanding that local-network device control will occur.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The activation description is broad and ambiguous, mixing unrelated use cases with actual device-control functionality. In an agent environment, vague invocation criteria can cause accidental or overbroad activation of a skill that has Bash, Read, Write, and Edit access plus the ability to affect devices on the network.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The skill includes automation and scheduled stop-all behavior that can interrupt displays or audio across multiple devices, but it does not provide a clear warning or consent boundary before performing network/device-impacting actions. In enterprise or shared environments, accidental execution could cause operational disruption or unauthorized control of presentation systems.

Static analysis

No suspicious patterns detected.