ClawHub
Back to skill

Security audit

GPT Image 2 多参考图生成

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent image-generation helper that uploads user-selected reference files to AI Hive and stores an API key for that service, with no evidence of hidden exfiltration or destructive behavior.

Install only if you are comfortable sending selected reference images and prompts to AI Hive and storing an AI Hive API key in your home directory. Avoid passing private or unrelated files to the upload/generate commands, and remove ~/.ai-hive/config.json if you no longer want the saved key.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill advertises and demonstrates shell execution, package installation, local file access, network access, and persistence of configuration, but does not declare those permissions. This creates a transparency and consent problem: users or hosting systems cannot accurately assess that the skill can read/write local data and send content to external services before running it.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding
The documented purpose is narrow—multi-reference image generation—but the skill behavior reportedly includes API key collection/storage, arbitrary media upload, browser launching, task querying, model/user/wallet introspection, and even chat/video capabilities. That mismatch is dangerous because it expands the attack surface beyond user expectations and could enable credential handling, data exfiltration, or unintended use of unrelated APIs under the guise of an image-generation skill.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.