Free Bilibili Ai Subtitle

• Outcome: Review
• Latest audit: May 29, 2026 at 9:33 PM UTC
• Version:
  1.0.0

Overview

This skill is not clearly malicious, but it sends videos and broad prompts to a third-party cloud video backend with scope and consent gaps.

Review this carefully before installing. Use it only with videos you are comfortable sending to NemoVideo, prefer a dedicated NEMO_TOKEN, and confirm uploads/exports before proceeding. Avoid private, confidential, biometric, copyrighted, or business-sensitive media unless you trust the provider and understand its retention and account policies.

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 (5)

Description-Behavior Mismatch

Medium

• Confidence: 94% confidence
• Finding: The manifest and top-level description frame the skill as a narrow Bilibili subtitle tool, but the documented behavior exposes a much broader cloud video editing, session, upload, state, and render workflow. This mismatch can mislead users and host systems about the skill’s actual capabilities, increasing the chance of overbroad routing, unintended data transfer, and use beyond the declared purpose.

Context-Inappropriate Capability

Low

• Confidence: 83% confidence
• Finding: The skill includes token handling, session creation, and credit-balance checks that go beyond a simple subtitle-generation flow. While these may support backend operation, exposing account-like operations without clearly tying them to the stated purpose creates unnecessary privilege surface and can normalize hidden service interactions.

Vague Triggers

High

• Confidence: 97% confidence
• Finding: Routing 'everything else' to the SSE backend makes the skill effectively a broad catch-all for arbitrary user prompts. This can cause the skill to activate outside its stated domain and forward unrelated user requests to a third-party cloud service, creating data exposure and prompt-scope expansion risks.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The invocation guidance includes the vague phrase 'Or just tell me what you're thinking,' which encourages broad, non-specific activation. In combination with remote backend processing, this increases the likelihood that unrelated user content or requests are sent to the service without clear intent or informed consent.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill asks users to drop video files in chat and emphasizes easy cloud processing, but it does not prominently warn that uploaded media and prompts are sent to a third-party backend. Because videos may contain sensitive personal, biometric, copyrighted, or confidential content, omission of clear disclosure undermines informed consent and materially raises privacy risk.

VirusTotal

66/66 vendors flagged this skill as clean.