ClawHub
Back to skill

Security audit

YouTube 提文案

Security checks across malware telemetry and agentic risk

Overview

This skill transparently extracts YouTube transcripts through disclosed third-party services, translates non-Chinese captions by default, and saves user-requested outputs locally.

Install only if you are comfortable sending YouTube links and transcript content to RedFox and, for non-Chinese captions, to Google Translate. Use --no-translate for original-language output and --no-save if you do not want transcript files written to Downloads.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill instructs use of environment variables, file output, and network/API access, but does not declare permissions or otherwise clearly surface those capabilities as required by the platform. This can mislead users and calling agents about what the skill will access, increasing the risk of unintended secret exposure, network egress, or local file writes.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The README instructs users to invoke the skill with very broad natural-language phrases such as asking to 'help extract the transcript' or 'export Excel'. In an agent environment, these generic phrases can overlap with ordinary conversation and cause accidental invocation or routing to this skill when the user did not explicitly intend it, potentially sending YouTube links and processing requests to external services unexpectedly.

Natural-Language Policy Violations

Medium
Confidence
84% confidence
Finding
The skill defaults to translating non-Chinese subtitles into Chinese and presenting Chinese output without first asking the user, which can cause unintended data disclosure to a third-party translation service and alter the original content. For transcript extraction, automatic transformation of user-requested content without opt-in is a policy and privacy risk, especially if the source material is sensitive or accuracy-critical.

Natural-Language Policy Violations

Medium
Confidence
90% confidence
Finding
The workflow section explicitly mandates showing the Chinese translated transcript and tells the agent to re-run with different flags only after delivery, removing the user's opportunity to choose language or avoid translation up front. This increases privacy risk, can degrade fidelity of the extracted transcript, and may send content to external translation infrastructure unnecessarily.

Natural-Language Policy Violations

Medium
Confidence
88% confidence
Finding
The result-display rules require Chinese transcript output for all users and define the transcript field as translated Chinese text, regardless of the source language. In context, this is more dangerous because the skill is designed to process arbitrary third-party video content, so forced translation can expose content externally and prevent users from receiving the exact source transcript they asked for.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.