妙达图片理解

Security checks across malware telemetry and agentic risk

Overview

This is a small image-analysis instruction skill that appears purpose-aligned, but users should treat submitted images as sensitive because it relies on an external AI CLI.

Install only if you trust miaoda-studio-cli and understand where it sends or processes images. Use it with images you intentionally choose, and avoid confidential screenshots, IDs, documents, or private photos unless that external AI processing is acceptable.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The trigger list is very broad and includes common phrases such as "看图", "图片识别", and "describe image", which can match ordinary user requests and cause the skill to activate when the user did not explicitly intend to invoke it. In this skill, unintended activation matters because it encourages sending a local image path to an external AI analysis command, potentially exposing sensitive local image content or causing the agent to choose the wrong tool.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill instructs users to pass a local image file to an AI-powered command but does not warn that the image content may be transmitted to an AI system for processing. This creates a privacy and consent risk: users may supply screenshots, photos, or documents containing sensitive information without understanding that the data leaves the local environment for remote analysis.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal