Hd Editor Ai

• Outcome: Review
• Latest audit: May 29, 2026, 9:32 PM
• Version:
  1.0.0

Overview

This appears to be a cloud video-editing skill, but it needs Review because it can automatically set up remote tokens/sessions and send prompts or media to a backend under broad triggers.

Install only if you are comfortable with prompts and selected media being processed by the remote video service, and prefer using it only after an explicit confirmation step. Avoid confidential footage unless you trust the provider, and monitor any NEMO_TOKEN, credits, or subscription effects.

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 (5)

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The skill is presented as simple HD video editing, but it also instructs the agent to obtain anonymous tokens, create backend sessions, and handle credit/subscription flows. That hidden expansion of scope can mislead users about account creation and backend interactions, increasing privacy and consent risk when media and identifiers are sent to a third-party service.

Description-Behavior Mismatch

Medium

• Confidence: 84% confidence
• Finding: The manifest claims support for only a small set of video formats, while the detailed behavior accepts many additional image, audio, and media types. This mismatch broadens the effective data intake beyond what users expect, which can cause unintended uploads or processing of different content types.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The invocation guidance is broad enough that ordinary editing or export phrases could activate the skill without the user clearly intending to use this specific remote video-processing service. In context, accidental activation is more dangerous because the skill can automatically connect to a backend and initiate account/session setup.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The routing rule sends 'Everything else' to the SSE editing action, creating an overly broad catch-all that can capture unrelated user prompts. Because that path transmits user text to a remote backend, an ambiguous request could result in unintended data disclosure or unexpected remote actions.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill instructs the agent to automatically connect to a remote backend and acquire an anonymous token on first use, while only telling the user 'Setting up...' and not clearly warning that media, prompts, and identifiers may be sent off-device. In a video-editing skill handling potentially sensitive footage, silent remote transfer and account/session creation materially increase privacy and consent risk.

VirusTotal

64/64 vendors flagged this skill as clean.