youtube-video-analyzer

Warn

Audited by Snyk on Feb 26, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.90). The skill explicitly downloads and ingests public YouTube content (video + subtitles) from a user-provided URL using yt-dlp (see SKILL.md Phase 1 and scripts/preprocess.sh) and the agent is instructed to read and act on subtitle text and frame images in Phase 3/4 prompts, so untrusted, user-generated third-party content could embed instructions that influence the agent's decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (high risk: 0.80). The preprocess.sh and other workflows download an arbitrary YOUTUBE_URL at runtime via yt-dlp (e.g., "https://youtube.com/watch?v=xxxxx"), extract subtitles/frames, and inject those remote subtitles and images directly into the agent prompts for analysis, so external content can control the model's inputs and behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 26, 2026, 08:53 AM