skills.rest
heman1033p-create
heman1033p-createCOMMUNITY
Explore all skills by this author →
GitHub

watch

Transcribes YouTube videos via Gemini and indexes transcripts into an Oracle knowledge base.

Updated Feb 28, 2026
Search Tags:#knowledge base#captions#youtube#transcription#yt-dlp#gemini#video learning
One-click install
npx skills add https://github.com/heman1033p-create/cv-v2-kongphop --skill watch-heman1033p-create
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: watch
Source: https://github.com/heman1033p-create/cv-v2-kongphop/tree/main/.gemini/skills/watch
Command: npx skills add https://github.com/heman1033p-create/cv-v2-kongphop --skill watch-heman1033p-create

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yt-dlp, mosquitto_pub, mosquitto_sub, and includes scripts (resource) components.

What problem does it solve? Manually watching and note-taking on YouTube videos is slow, and transcripts get lost without a system. This Skill automates sending YouTube videos to Gemini for transcription, then saves and indexes the results into a persistent knowledge base for later retrieval. ## Core Features & Use Cases - Gemini Transcription: Sends YouTube URLs to Gemini with mode options (chat, research, canvas) and model selection (fast, thinking, pro) via MQTT browser automation. - Metadata & Captions: Fetches video title, channel, duration, and YouTube auto-captions in SRT format using yt-dlp. - Knowledge Persistence: Saves transcripts as dated Markdown learning files with frontmatter and indexes key takeaways into Oracle for semantic search. - Use Case: You find a 45-minute conference talk on YouTube. Run the watch command to get a timestamped Gemini transcript, cross-checked against YouTube captions, saved to your learnings folder, and searchable later via Oracle. ## Quick Start Ask the assistant to watch and transcribe a YouTube video by providing its URL, for example by saying to watch https://youtube.com/watch?v=xxx and save the key takeaways.

Frequently Asked Questions about watch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I transcribe a YouTube video with Gemini?▼

Run the transcribe script with a YouTube URL, and it fetches metadata via yt-dlp, opens a Gemini tab through MQTT browser commands, and sends a transcription prompt. The transcript appears in the Gemini tab within 30-60 seconds for long videos.

How to get YouTube captions in SRT format from the command line?▼

Use the get-cc script with a video URL and optional language code. It calls yt-dlp with --write-auto-sub and --sub-format srt, then prints the caption file contents to stdout, or outputs NO_CAPTIONS_AVAILABLE if none exist.

What Gemini modes are supported for video transcription?▼

Three modes are supported: chat for quick transcription, research for deep analysis with fact-checking, and canvas for structured document output. You can also select fast, thinking, or pro models via the --model flag.

Does YouTube transcription work if Gemini cannot access the video?▼

If Gemini cannot access the video, it will say so explicitly, and the fallback is to save the URL with manual notes. The user must also be logged into Google, otherwise Gemini blocks the request.

Why does the transcription script fail to create a Gemini tab?▼

Tab creation depends on an MQTT broker at localhost:1883 relaying commands to a browser extension. If no tabId response arrives within the timeout, the script continues with a warning, but prompt delivery may fail without the browser bridge running.

Related Skills

Explore More Education & Research Skills→
MSApps-MobileMSApps-Mobile
8

youtube-transcriber

Extract YouTube video and playlist transcripts with timestamps from browser captions.

Official
Advanced
gh-xjgh-xj

study-assistant

Extract key concepts and answer questions from course transcripts and notes.

Community
Basic
thiansitthiansit

watch

Transcribe YouTube videos into searchable transcripts and knowledge entries.

Community
Intermediate
enclawedenclawed
2

openai-whisper

Transcribe audio files to text locally using the Whisper CLI.

Community
Basic