Free Video Generator Hugging Face

• Outcome: Review
• Latest audit: May 29, 2026 at 9:35 PM UTC
• Version:
  1.0.0

Overview

This skill may generate videos, but it is misleadingly branded around Hugging Face while automatically sending prompts, media, and session data to NemoVideo's cloud API.

Review before installing. Use only if you are comfortable sending prompts, images, video, audio, URLs, and render requests to NemoVideo's cloud service, and do not assume Hugging Face or local open-source processing. Avoid confidential media and protect or rotate any NEMO_TOKEN used with it.

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 top-level description frame this as a simple prompt-to-video generator, but the body grants a much broader remote editing/export capability with session state inspection, timeline manipulation, uploads, downloads, audio/text overlays, and multi-format media handling. This scope mismatch is dangerous because users and platform reviewers may consent to a narrow function while the skill actually routes arbitrary content and editing actions to a third-party cloud service, increasing the chance of unauthorized data handling or abuse.

Context-Inappropriate Capability

Medium

• Confidence: 92% confidence
• Finding: The skill automatically performs token acquisition, session creation, and exposes credits/balance and export-billing semantics that are not clearly necessary for a user-facing 'free video generator' workflow. This is risky because it normalizes silent account/session operations against an external service and can obscure when a user's actions create persistent sessions, consume quotas, or interact with monetized features.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The catch-all routing rule sends essentially any unmatched user input into the SSE-backed remote action pipeline. This is dangerous because unrelated, sensitive, or ambiguous user text may be forwarded to the third-party backend without clear intent validation, causing over-collection of user data and unintended remote operations.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill instructs the agent to automatically connect to an external API and, if needed, obtain an anonymous token on first interaction before doing anything else, without first warning the user that prompts, files, and metadata may be transmitted off-platform. In this context, the skill handles uploads and free-form prompts, so silent network setup materially increases privacy and consent risk and could expose user content to a third-party processor unexpectedly.

VirusTotal

65/65 vendors flagged this skill as clean.