Video Narration Ai

• Outcome: Review
• Latest audit: May 28, 2026, 8:55 AM
• Version:
  1.0.0

Overview

This skill is a real cloud video narration/editing helper, but it can broadly send video files and ambiguous editing prompts to a third-party service without clear consent boundaries.

Review before installing. Use only for videos you are comfortable sending to NemoVideo's cloud service, and require the agent to ask before creating sessions, uploading files, sending broad edit prompts, exporting renders, or spending credits.

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 (4)

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The manifest and marketing position this as a voiceover-narration tool, but the documented behavior exposes a much broader remote video editing pipeline including uploads, state management, chat-driven edits, and export/render workflows. This scope mismatch can cause overbroad invocation and user consent problems because users may believe they are using a narrow narration feature while the skill can perform substantially wider media manipulation and data transmission.

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: The skill advertises narration, yet the routing and examples cover generic editing actions such as aspect ratio changes, overlays, audio-track manipulation, and broad media operations. That unjustified expansion increases the chance of unintended invocation and remote processing of user media for actions outside the user's expected consent boundary.

Vague Triggers

Medium

• Confidence: 97% confidence
• Finding: The catch-all rule routes essentially all unmatched requests to the SSE editing action, making the skill eligible for a very wide range of prompts beyond its stated purpose. Overbroad routing can cause accidental activation, unintended remote API calls, and unexpected transmission of user content or instructions to a third-party service.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill tells users to upload video files for remote processing, but it does not provide a clear user-facing warning that media will be transmitted to an external service and processed server-side. This creates a privacy and consent risk, especially for sensitive or personal videos, because users may not understand where their files are going or what retention/handling applies.

VirusTotal

VirusTotal findings are pending for this skill version.