YouTube Thumbnail Generator with Nano Banana

• Outcome: Pass
• Latest audit: May 29, 2026, 8:16 PM
• Version:
  0.1.1

Overview

This skill coherently uses Gemini to create YouTube thumbnail plans and optional images, with expected local output files and no evidence of hidden or destructive behavior.

Install this only if you are comfortable using a Gemini or Google API key and sending thumbnail titles, scripts, hooks, or marketing copy to Gemini. Keep generated outputs in a dedicated directory, avoid prompts containing secrets or highly confidential unpublished material, and review metadata JSON before sharing or committing it.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 93% confidence
• Finding: The skill instructs the agent to execute local Python scripts, consume API keys from the environment, write files under outputs/, and call external Gemini endpoints, yet it does not declare corresponding permissions or provide an explicit capability boundary. This creates a trust gap where a caller may treat the skill as low-risk documentation while it actually performs shell, network, environment, and filesystem operations.

Missing User Warnings

Low

• Confidence: 84% confidence
• Finding: The skill directs writing JSON and image artifacts to local paths and saving sidecar metadata, but it does not warn that running the workflow will modify the local filesystem. In agent environments, undisclosed writes can overwrite files, leak prompts or metadata into workspace artifacts, or violate user expectations about read-only analysis.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill states that scripts use GEMINI_API_KEY/GOOGLE_API_KEY and call official Gemini image/text endpoints, but it does not clearly disclose that user-provided copy and prompts may be transmitted to a third-party service. That omission can expose sensitive marketing drafts or proprietary content to external processing without informed consent.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The script sends user-provided copy, audience, style, and related prompt context to an external Gemini API, but the code provides no explicit notice, consent flow, or redaction guard before transmission. This can expose proprietary marketing copy, unpublished campaign details, or other sensitive text to a third-party service, which is a real privacy and data-handling risk even if it is expected behavior for the feature.

VirusTotal

65/65 vendors flagged this skill as clean.