Photo Video Maker Best

• Outcome: Pass
• Latest audit: May 29, 2026, 8:37 PM
• Version:
  1.0.0

Overview

This is a coherent cloud media-editing skill that uploads user-selected media and prompts to NemoVideo for rendering, with privacy caveats but no artifact-backed malicious behavior.

Install only if you are comfortable sending selected photos, videos, audio, prompts, and timeline data to NemoVideo for cloud processing. Avoid sensitive media unless you trust that provider's privacy and retention practices, and watch credit or subscription usage before exporting.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The skill is presented as a photo-to-video tool, but the documented accepted formats and timeline/export workflow materially expand its capabilities into general media editing with audio/video ingestion. That scope mismatch can mislead users and host platforms about what data will be handled and what actions the skill may take, increasing the chance of unintended upload of sensitive media and broader-than-expected remote processing.

Context-Inappropriate Capability

Low

• Confidence: 78% confidence
• Finding: The skill instructs runtime detection of local install paths and references a local config directory for attribution/platform detection, which is unnecessary for core photo/video generation. Accessing environment/layout details beyond functional need expands local information exposure and can leak host-specific metadata to a remote service through headers or telemetry.

Vague Triggers

Medium

• Confidence: 81% confidence
• Finding: The catch-all routing rule sends 'everything else' to the SSE backend, meaning a very wide range of user inputs may be transmitted to a remote service without sufficiently specific intent matching. In this skill context, that increases the risk of over-collection and unintended disclosure of user prompts, especially because the backend appears to interpret free-form editing commands and GUI-like actions.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill automatically connects to external endpoints, acquires tokens, creates sessions, and uploads user files/prompts to a remote backend, but it does not provide a clear user-facing warning or consent step before network transmission. Because the tool handles personal photos, images, and potentially other media, the absence of explicit disclosure materially increases privacy and data-handling risk.

Natural-Language Policy Violations

Medium

• Confidence: 74% confidence
• Finding: Hard-coding the session language to English without user choice can cause prompts or backend interpretations to be translated or processed in an unintended language, potentially altering meaning of user instructions and metadata. In a media-processing workflow this is mainly a privacy/integrity issue, since user content may be misinterpreted or unnecessarily normalized before being sent to the backend.

VirusTotal

66/66 vendors flagged this skill as clean.