custom-youtube-summarize

What to consider before installing/using: - Functional bug: The script's start marker is wrong. SKILL.md expects '[TRANSCRIPT-START]' but transcript_extract.py prints '[TRANSCRIPT-START\n'. Either fix the script to print '[TRANSCRIPT-START]\n'+text+'\n[TRANSCRIPT-END]' or update the capture logic to match the actual output. Without this fix the agent may not extract the transcript correctly. - Network activity: The script performs outbound requests to YouTube to fetch transcripts. This is expected for the feature but be aware these requests may be rate-limited or blocked; the bundled library suggests using proxies (and has optional flags for proxy username/password) if you hit IP blocking. - No secrets requested: The skill does not ask for environment variables or keys. Only provide proxy credentials if you intentionally configure that feature and trust the proxy provider. - Code provenance & maintenance: The skill bundles a third-party library (youtube_transcript_api) rather than declaring it as a dependency. That increases the code you should review for licensing and timely security fixes. If you prefer, replace the bundled library with a vetted package from a package manager and pin a known-good version. - Test before enabling: Run the included script locally with a known YouTube URL to verify behavior and ensure the transcript markers are produced as your agent's capture logic expects. If you want, I can produce a small fixed version of transcript_extract.py (correcting the marker and optionally adding a --format or timeout) or a minimal validation checklist to run before enabling this skill.