Create a verbatim transcript for a YouTube URL using Google Gemini (speaker labels, paragraph breaks; no time codes). Use when the user asks to transcribe a YouTube video or wants a clean transcript (no timestamps).
Security Analysis
high confidenceThe skill's code, requirements, and runtime instructions align with its stated purpose (transcribing YouTube videos via Google Gemini) and do not request unrelated credentials or perform unexplained actions.
Name/description match the actual behavior: the script calls Google Gemini to generate a transcript from a YouTube URL. Declared requirements (python3, GEMINI_API_KEY) are appropriate for that functionality.
SKILL.md instructs running the included Python script which: fetches the YouTube title via oEmbed, sends a prompt and the video URL to the Gemini API (using a 'file_data' file_uri so Gemini can ingest the video), and writes a transcript file under the workspace or /tmp. The actions are within scope for transcription, but the script explicitly sends the video URL (and relies on Gemini fetching the video content) to Google — a privacy/third-party data-sharing consideration rather than an incoherence.
No install spec; instruction-only with a small Python script that uses only the standard library (urllib, json, pathlib). No external downloads or archives, so installation risk is low.
Only GEMINI_API_KEY is required. The script optionally reads OPENCLAW_WORKSPACE or TMPDIR for safe output path resolution; no unrelated secrets or excessive env variables are requested.
The skill is not always-enabled, does not modify other skills or global agent settings, and only writes transcript files into controlled workspace or /tmp locations. No elevated persistence is requested.
Guidance
This skill appears coherent for its stated purpose, but consider privacy and key-handling before installing: the script sends the video URL (and relies on Gemini fetching the video content) to Google’s generative-language API using your GEMINI_API_KEY, so private or DRM-protected videos may be exposed to Google and may fail to be ingested. Only provide a GEMINI_API_KEY you trust this use for; run the script in a controlled workspace; examine the code (it’s small and uses only standard Python libs); and if the key is later compromised, revoke/rotate it. If you need transcripts without sending media to Google, this skill is not suitable.
Latest Release
v1.0.4
Fix metadata: moltbot→openclaw so registry picks up GEMINI_API_KEY and python3 requirements
More by @odrobnik
Published by @odrobnik on ClawHub
