ClawHub
Back to skill

Security audit

爆款视频字幕与文案拆解｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE workflow helper for Chinese short-video caption, copy, and generation tasks, with expected but sensitive API-key, upload, download, and local ffmpeg behavior.

Install only if you intend to use AI-HIVE for short-video caption/copy workflows and media generation. Review commands before running them, do not paste API keys into chat, confirm any paid generation parameters first, and upload only media you are authorized to use.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill documents code paths that use environment variables, filesystem access, network calls, and shell-based tooling, but it does not declare permissions or clearly constrain those capabilities. This creates a trust and review gap: a caller may invoke the skill expecting only text analysis, while the skill can read local files, invoke ffmpeg, access secrets from the environment, and make outbound requests.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The declared purpose is subtitle/copy analysis, but the skill also exposes broad AI-HIVE chat, image/video generation, media upload/download, local video processing, and setup flows. That mismatch is dangerous because it can smuggle in far more powerful behavior than users or reviewers expect, increasing the chance of unauthorized file handling, external data transfer, cost-incurring actions, or misuse of local tooling.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The manifest advertises a very broad default prompt and also enables implicit invocation, which can cause the skill to trigger for loosely related requests without clear user intent. In a skill that generates workflows, runnable commands, and production-ready marketing outputs, overbroad activation increases the chance of unintended execution paths, unsafe content generation, or bypass of user expectations and policy boundaries.

Natural-Language Policy Violations

Medium
Confidence
80% confidence
Finding
The default prompt hard-codes Chinese output without any user opt-in, which can override user preference and reduce transparency about how the assistant will respond. While not directly enabling code execution or data exfiltration, forced language behavior can mislead users, degrade reviewability, and make unsafe generated commands or policy-sensitive content harder for some users to evaluate before use.

Static analysis

No suspicious patterns detected.