Text To Video Meta Ai

• Outcome: Review
• Latest audit: May 29, 2026 at 9:34 PM UTC
• Version:
  1.0.0

Overview

This looks like a cloud video-generation skill, but its instructions are broader and less clearly disclosed than its text-to-video description suggests.

Review before installing. Use it only if you are comfortable sending prompts, uploaded files, and possibly media assets to the provider's cloud video service. Avoid confidential documents or private media unless you trust the provider, and require explicit confirmation before uploads, generation, exports, or use of any service token.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The skill metadata and description present this as a narrow text-to-video tool, but the body documents a much broader remote video-editing capability including uploads, audio/text overlays, state inspection, and export workflows. This scope mismatch can mislead users and host agents about what data and actions are actually permitted, increasing the risk of unintended data exposure or overbroad invocation.

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The manifest says uploads are limited to text-oriented formats, but the documented behavior supports a much wider set of media formats for upload/export. This discrepancy can cause users or orchestrators to share files under false assumptions about the tool's scope, resulting in broader third-party data transfer than advertised.

Intent-Code Divergence

Low

• Confidence: 81% confidence
• Finding: The later guidance expands usage beyond uploaded text files into broader media workflows, which conflicts with the earlier stated input scope. While this is primarily a transparency and expectation-setting issue, it can still contribute to unsafe data handling by encouraging users to submit content they did not expect to leave their environment.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The activation phrase guidance is broad enough to match common conversational language such as generic requests to generate prompts. Overly permissive triggering can cause unintended skill activation and remote transmission of user text to the backend without clear intent to use this specific service.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: Routing 'everything else' to generation creates an overly broad catch-all that can treat unrelated user input as permission to send content to the remote generation backend. This increases the chance of accidental invocation, surprising data disclosure, and actions outside the user's intended scope.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The top-level description markets the functionality but does not clearly warn that prompts and uploaded files are sent to a remote cloud backend. Because the skill processes potentially sensitive user content, omission of this disclosure undermines informed consent and can lead to unintended exfiltration of private documents or media.

VirusTotal

64/64 vendors flagged this skill as clean.