Security audit

Nano Banana Pro 广告图片

Security checks across malware telemetry and agentic risk

Overview

This is an AI Hive image-generation skill that uses user-supplied prompts and media, with some credential and upload behavior users should understand before use.

Install only if you are comfortable using an AI Hive API key and uploading selected image/reference assets to AI Hive. Use environment variables or the local config deliberately, keep the API key private, and do not pass sensitive unrelated files to the upload or image options.

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 (4)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 86% confidence
Finding: The skill invokes shell commands, installs packages, accesses local files, and uses networked Python scripts, but the manifest does not declare any permissions or capability boundaries. This creates a trust and review gap: users may approve an apparently simple image skill without understanding it can read/write files, access environment data, and make outbound requests.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The documented purpose is limited to ad-image generation/editing, but the underlying behavior reportedly includes broader capabilities such as chat, video generation, arbitrary media upload, account information retrieval, model discovery, API key setup, and local credential storage. This mismatch is dangerous because it can mislead users into granting access to data and operations far beyond the advertised scope, increasing the risk of privacy loss, credential exposure, or unintended account actions.

Description-Behavior Mismatch

Medium

Confidence: 87% confidence
Finding: The skill is presented as a narrowly scoped ad-image generator, but it also exposes generic task inspection and arbitrary media upload commands. This broadens the operational surface beyond user expectations and can enable unintended data movement or access to task metadata unrelated to the advertised purpose.

Context-Inappropriate Capability

Medium

Confidence: 84% confidence
Finding: The skill contains an API-key bootstrap flow that opens a browser and guides the user through account login and key creation, which exceeds the expected behavior of a simple image-generation skill. In a hostile or misleading skill context, this kind of credential acquisition workflow can normalize secret collection and increase phishing or over-privileged account linking risk.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.