AgentSkillsCN

universal-transcriptor

利用 Google Gemini 1.5 Flash 为 YouTube 视频生成逐字转录稿。当用户要求对 YouTube 视频进行转录、获取转录文本或提取视频中的文字时,可选用此功能。直接支持 YouTube 链接,无需下载。输出经过格式化处理的转录稿,并按主题进行分段。触发词:转录、转写、获取转录、YouTube 转录、视频转文字。

SKILL.md
--- frontmatter
name: universal-transcriptor
description: >
  Generate verbatim transcripts for YouTube videos using Google Gemini 1.5 Flash.
  Use this when the user asks to transcribe, get transcript, or extract text from YouTube videos.
  Works directly with YouTube URLs - no downloading required. Outputs formatted transcripts with topic segmentation.
  Triggers: transcript, transcribe, get transcript, youtube transcript, video to text
version: 1.0.0
homepage: https://github.com/prayag001/Universal-Transcription
metadata:
  openclaw:
    emoji: "📝"
    requires:
      env: ["GEMINI_API_KEY"]
      bins: ["python"]

Universal Transcriptor

Generate complete, verbatim transcripts for YouTube videos using Google Gemini 1.5 Flash.

⚠️ WHEN TO USE

  • When the user asks to transcribe a YouTube video
  • When the user wants the full text content from a video
  • When the user provides a YouTube URL and asks for a transcript

How to Use

Step 1: Get the YouTube URL

Identify the YouTube URL from the user's request (e.g., https://youtu.be/... or https://www.youtube.com/watch?v=...)

Step 2: Run the script

bash
python "{baseDir}/scripts/universal_transcript.py" "<YOUTUBE_URL>"

Example:

bash
python "{baseDir}/scripts/universal_transcript.py" "https://youtu.be/dQw4w9WgXcQ"

Step 3: Present the results

The script will:

  1. Generate a complete verbatim transcript
  2. Save it to ~/.openclaw/workspace/out/<video-title>.txt
  3. Print the file path and a preview
  4. Show the first 15 lines as a preview

Present the saved file path and key content to the user.

Output Format

  • Verbatim transcript with topic segmentation
  • Timestamps at topic changes (format: ### [MM:SS] Topic: <description>)
  • Vertical formatting (short lines for readability)
  • Includes metadata header (title, URL, date)
  • Saved to workspace/out/ directory

Requirements

  • GEMINI_API_KEY must be set in environment or in ~/.openclaw/.env
  • Python 3 installed and accessible