Nano Banana Pro Prompts

• Outcome: Review
• Latest audit: May 29, 2026 at 9:28 PM UTC
• Version:
  1.1.1

Overview

The skill is mainly a prompt recommender, but it auto-refreshes mutable remote data and requires fetching or sending external preview images, so it needs user review before installation.

Install only if you are comfortable with the agent fetching prompt data from GitHub, refreshing local reference files over time, and accessing external preview-image URLs. Prefer reviewing scripts/setup.js first and using manual updates if you need stable, auditable prompt data.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• 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
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The skill tells the agent to execute shell commands to download remote files and then delete local files as part of prompt recommendation. That exceeds the core need of a search/recommendation skill and introduces unnecessary command execution and file-system side effects based on untrusted `sourceMedia` data.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The package description explicitly states that the skill auto-downloads a library on install, which is a supply-chain and user-consent risk because installation triggers network activity before the user has a chance to review or approve it. In an agent-skill context, install-time downloads can fetch mutable remote content and expand the trust boundary beyond the published package.

VirusTotal

59/59 vendors flagged this skill as clean.