Ai Image To Video Deepfake

• Outcome: Pass
• Latest audit: May 29, 2026, 8:38 PM
• Version:
  1.0.0

Overview

This is a disclosed cloud deepfake/video-generation skill, but users should be cautious about sending personal media to the external service.

Install only if you trust the NemoVideo backend and are comfortable sending selected media and prompts to it. Avoid uploading sensitive likenesses, voices, private images, or third-party content without permission, keep NEMO_TOKEN private, and label realistic generated videos as synthetic when sharing them.

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
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The suggested trigger phrases are broad enough to activate on common conversational input, increasing the chance that the skill engages unexpectedly. In a skill that uploads user media to a third-party cloud backend and performs deepfake generation, accidental activation can lead to unintended disclosure of personal files or initiation of sensitive processing without clear user intent.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The catch-all routing rule sends 'Everything else' to the SSE action, which makes the skill prone to processing arbitrary user input as backend commands. Because SSE can drive editing actions and remote processing, ambiguous routing broadens the attack surface and can cause unintended requests, state changes, or media handling based on unrelated or unclear prompts.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill does not clearly warn users that their images, audio, and related media are sent to a cloud service for processing. Given that the skill explicitly handles personal portraits and deepfake generation, omission of a prominent disclosure materially increases privacy and consent risk, especially for sensitive biometric-like face data and potentially copyrighted or intimate media.

VirusTotal

60/60 vendors flagged this skill as clean.