youtube-extract

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires verbatim extraction of links/code/command invocations from transcripts (with timestamps) so any API keys, tokens, cookies, or passwords spoken in the video would be output unchanged, creating an exfiltration risk.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most of the links are benign (YouTube, Stripe, Anthropic org, example.com), but the placeholder "https://{raw}" and the GitHub repo at github.com/anthropics/claude-code (likely a typo/typosquat of Anthropic/Claude) are high-risk indicators for malicious distribution, so treat the set as moderately to highly suspicious overall.

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

• Third-party content exposure detected (high risk: 0.95). Outsider free text enters the LLM context via the runtime-fetched YouTube transcript segments (scripts/fetch.py downloads captions from YouTube and returns transcript[] text, which is then passed into the Pass 2 LLM prompt prompts/extract_artifacts.md).