ClawHub
Back to skill

Security audit

GPT Image 2 商品精修

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed AI Hive product-photo retouching helper that uploads user-selected images, runs a fixed image model, and saves results without hidden persistence or destructive behavior.

Install only if you are comfortable sending the selected product images and prompts to AI Hive or a base URL you configure. Keep the stored API key protected, use --param only when you know what the backend parameter does, and review outputs against the original image when product accuracy matters.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill invokes shell commands, reads user-supplied image paths, writes configuration and output files, and performs network requests, yet it declares no permissions or capability boundaries. This creates a transparency and control gap: users or hosting platforms cannot accurately assess or restrict what the skill can access, increasing the risk of unintended file access, data exfiltration, or unsafe execution in more privileged environments.

Description-Behavior Mismatch

Medium
Confidence
83% confidence
Finding
The skill is marketed as a constrained product-retouching tool, but the exposed `generate` command accepts an arbitrary user prompt and forwards it directly to a general image-generation endpoint using a fixed image model. This scope mismatch weakens policy and operational controls because users can repurpose the skill for broader image synthesis beyond the stated business purpose, which may bypass governance, moderation expectations, or usage restrictions tied to retouch-only workflows.

Context-Inappropriate Capability

Medium
Confidence
87% confidence
Finding
The code exposes arbitrary `--param key=value` passthrough and forwards the resulting dictionary directly into the backend `params` field without allowlisting. That enables callers to invoke undocumented or unsafe model behaviors, alter safety-relevant settings, or access capabilities inconsistent with the skill's declared retouching purpose.

VirusTotal

56/56 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.