ClawHub
Back to skill

Security audit

GPT-image2

Security checks across malware telemetry and agentic risk

Overview

This appears to be a normal image-generation skill that uses Redfox for processing, though users should notice that Redfox, not a direct OpenAI API key, is the service endpoint shown in the artifacts.

Install only if you are comfortable using Redfox as the API provider. Treat prompts, selected reference images, and the Redfox API key as data sent to an external service; avoid private photos, confidential designs, regulated data, or secrets unless you have reviewed Redfox's privacy and retention terms. Prefer an environment variable or a protected config file for the API key, and rotate the key if it is exposed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
97% confidence
Finding
The manifest and description present the skill as an OpenAI gpt-image-2 integration, but the setup instructions require users to register with and authenticate to Redfox instead. This mismatch can mislead users about who receives their prompts, images, and API key material, undermining informed consent and potentially redirecting sensitive data to an unexpected third party.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The documentation repeatedly claims calls go to OpenAI's latest image model, while the operational instructions indicate use of Redfox credentials and service endpoints. That inconsistency is dangerous because it can cause users to trust the wrong party with uploaded reference images, generated content, and billing or credential information.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The examples encourage users to upload local images and provide API keys, but they do not clearly warn that prompts and reference images will be transmitted to an external service. In this skill context, that omission is more dangerous because users may unknowingly send personal photos or sensitive visual data off-device to a third party.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.