ClawHub
Back to skill

Security audit

商品图转UGC带货视频｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE workflow for turning product images into marketing images/videos, with expected API, media upload, local output, and ffmpeg helper behavior.

Install only if you intend to use AI-HIVE for product image/video generation. Use authorized product media, review prompts and routing before submitting billable jobs, and avoid placing real API keys in logs, screenshots, or committed files. If you use the init command, it stores your AI-HIVE key locally in ~/.ai-hive/config.json.

SkillSpector

Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding
The skill advertises executable workflows that use environment variables, local file access, shell commands, network calls, and ffmpeg, but no explicit permission declaration is present. This creates a trust and review gap: consumers may invoke code-capable behavior without clear consent boundaries, increasing the risk of unintended data access, secret exposure, or command execution in agent environments.

Tp4

High
Category
MCP Tool Poisoning
Confidence
89% confidence
Finding
The documented purpose is a narrowly scoped product-image-to-UGC-video workflow, but the described behavior appears to include broader capabilities such as generic chat/model access, wallet retrieval, standalone uploads, broader image editing, ffmpeg operations, and interactive API-key setup. This mismatch is dangerous because it can mask overbroad functionality from users and reviewers, enabling unnecessary access to account metadata, media, and local execution paths beyond the stated use case.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The skill enables implicit invocation without any visible trigger constraints, allowing the system to auto-select this workflow for loosely related requests. Because this skill can generate runnable commands and production workflows for external API usage, overly broad auto-invocation increases the chance of unintended execution paths, user confusion, or application in disallowed marketing/deception scenarios.

Natural-Language Policy Violations

Medium
Confidence
81% confidence
Finding
The default prompt hard-codes Chinese output regardless of the user's language preference, which can cause mismatches between user intent and system behavior. While this is not a direct security exploit, it can reduce transparency and informed consent, especially if operational instructions or policy-relevant warnings are delivered in a language the user did not request.

Static analysis

No suspicious patterns detected.