ClawHub
Back to skill

Security audit

投屏设备控制

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a Chromecast control guide, but it asks for shell/LAN control while its scope and dependency disclosures are too loose for automatic installation without review.

Review before installing. Use this only for explicit Chromecast or catt tasks, and be aware it may run local shell commands, discover/control devices on your LAN, cast selected local files or URLs to external devices, and modify catt configuration under your home directory. Avoid using it on public or untrusted networks, and confirm local file paths and target devices before casting.

SkillSpector

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)

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The skill metadata and description mix unrelated claims about video/audio processing, media conversion, and dubbing with Chromecast control behavior. This mismatch can mislead users and downstream agents about what the skill actually does, increasing the chance that it is invoked in inappropriate contexts or granted permissions under false assumptions.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The dependency section omits the required external `catt` executable and local command-execution behavior, claiming only an LLM API is needed. This hides the real trust boundary: the skill requires shell execution and network/device control, which could cause users or platforms to enable it without understanding its operational and security implications.

Vague Triggers

Medium
Confidence
84% confidence
Finding
Overly broad invocation language unrelated to Chromecast control makes the skill eligible for prompts outside its intended scope. In an agent ecosystem, ambiguous routing can cause accidental execution of a privileged skill with `exec`, `read`, and `write` capabilities when a less-privileged tool should have been used.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill documents casting local files and persisting aliases/default-device settings in `$HOME/.config/catt/catt.cfg` without clearly warning that local content may be exposed over the network and that host configuration will be modified. In this context, the skill directly interacts with LAN devices and local files, so missing disclosure increases the risk of unintended data exposure and persistent system changes.

Static analysis

No suspicious patterns detected.