Gemini Image CLI

• Outcome: Pass
• Latest audit: May 29, 2026, 8:40 PM
• Version:
  0.1.0

Overview

This is a disclosed Gemini image-generation helper that sends chosen prompts and images to Gemini or a configured proxy and saves results locally.

Install only if you are comfortable sending your prompts and any selected reference images to Google Gemini or to the configured proxy. Use --provider local for more sensitive work, keep API keys scoped and private, and avoid retries, batches, or --overwrite unless you intend the extra quota use or file replacement.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (2)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 91% confidence
• Finding: The skill instructs the agent to invoke a bundled shell script that performs outbound network requests, yet the skill metadata declares no permissions. This mismatch can bypass user or platform expectations about what the skill is allowed to do, increasing the risk of unintended command execution and data egress, especially because the workflow encourages direct script execution and provider auto-selection.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The description is broad enough to match generic troubleshooting, debugging, and request-inspection tasks, which can cause the skill to activate outside narrowly intended image-generation use cases. In context, that is riskier because the skill has shell and network capabilities and could be invoked for loosely related requests, leading to unnecessary external calls, script execution, or exposure of raw request/response data.

VirusTotal

66/66 vendors flagged this skill as clean.