Trimmer Software

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

Overview

This appears to be a real cloud video-trimming skill, but it can start third-party sessions and route broad user prompts to Nemo’s backend without clear consent or scope limits.

Install only if you are comfortable sending video, audio, images, URLs, editing prompts, and session metadata to nemovideo.ai. Use a token intended only for this service, avoid sensitive recordings unless you trust the provider’s data handling, and confirm before allowing anonymous session creation or processing ambiguous requests.

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
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

• Confidence: 91% confidence
• Finding: The skill instructs the agent to automatically obtain anonymous tokens and create backend sessions with free credits, which expands behavior beyond simple local trimming into account-like resource acquisition on a third-party service. This can consume service credits without explicit user consent, hide the fact that the agent is authenticating to a remote backend, and normalize unattended access to billable or rate-limited resources.

Context-Inappropriate Capability

Medium

• Confidence: 84% confidence
• Finding: The skill includes credit and balance checking capabilities that are not necessary for the core trimming workflow. Exposing account-resource management broadens the action surface and may encourage the agent to inspect or act on billing-related state without a clear user need.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The prompt guidance 'Or just tell me what you're thinking' is overly broad and can cause accidental invocation from ordinary conversation. In a skill that uploads media and communicates with a cloud backend, ambiguous activation increases the chance of unintended data transfer or backend actions.

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The routing rule that sends 'Everything else' to the SSE backend gives the skill effectively open-ended scope. That allows unrelated user input to be forwarded to a remote service, increasing the risk of privacy leakage, unintended operations, and behavior outside the stated trimming purpose.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill sends prompts, uploaded files, and session metadata to a third-party cloud backend but does not present a clear user-facing warning before doing so. For a media-processing skill handling potentially sensitive recordings, lack of disclosure materially increases privacy and consent risk.

VirusTotal

61/61 vendors flagged this skill as clean.