Video Subtitle Extractor

• Outcome: Review
• Latest audit: May 28, 2026 · 9:02 AM UTC
• Version: 2.0.0

Overview

This appears to be a real subtitle extraction tool, but it needs review because it recommends browser-cookie based downloads without adequate account/session privacy warnings.

Install only if you are comfortable with a tool that downloads media, installs local ASR dependencies, downloads models, and stores audio, optional video, transcripts, and metadata locally. Use a dedicated output directory, avoid full-video saving unless needed, and do not use browser-cookie options unless you understand they can expose logged-in account sessions; prefer a dedicated browser profile and delete downloaded artifacts or cookie exports when finished.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

if isinstance(cmd, list):
            subprocess.run(cmd, check=check, capture_output=False)
        else:
            subprocess.run(cmd, shell=True, check=check, capture_output=False)
        return True
    except subprocess.CalledProcessError:
        return False

• Confidence: 96% confidence
• Finding: subprocess.run(cmd, shell=True, check=check, capture_output=False)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 90% confidence
• Finding: The skill advertises and documents shell execution, environment access, and file read/write behavior, yet no permissions are declared. That creates a trust and enforcement gap: an agent or reviewer may assume the skill is lower risk than it really is, while the documented workflow clearly invokes local commands, installs packages, downloads files, and persists artifacts.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 84% confidence
• Finding: The documented behavior materially exceeds or differs from the stated description: it can download full video files, persist multiple artifacts, rely on rule-based calibration rather than purely LLM-based processing, and requires some manual model setup despite claiming automatic setup. These mismatches can mislead users and security controls about what the skill actually does, increasing the chance of unsafe execution or improper consent.

Missing User Warnings

Low

• Confidence: 77% confidence
• Finding: The skill persists downloaded media, transcripts, metadata, and calibrated outputs to disk, but the top-level description does not prominently warn users about storage consumption and local data retention. This is primarily a transparency and privacy-risk issue rather than direct exploitation, but it can still lead to unanticipated retention of potentially sensitive media or transcripts.

Missing User Warnings

Medium

• Confidence: 86% confidence
• Finding: The documentation recommends using browser-cookie extraction for authenticated downloads without an explicit privacy and security warning. Browser cookies can grant account access and may expose session data to tools, logs, or local storage if mishandled, so omitting a warning increases the risk of credential or session misuse.

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.