video-content-watcher

针对 YouTube、Bilibili 或本地视频文件，先用 yt-dlp 或 ffmpeg 抽取音频，再用 Whisper 转写，最后调用 MiniMax 大模型生成含基本信息的 Markdown 分析报告，涵盖核心摘要、关键要点、主要观点、优缺点与目标受众。

franklinyung

热度 39.75

最近更新 2026-05-11 17:35:46

分类：数据依赖：代码执行 / 读写文件 / 第三方 API / 内部调用大模型应用场景：摘要总结 / 报告撰写 / 研究

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add video-content-watcher

Metadata

name: video-content-watcher
description: Extract core content and generate structured analysis reports from YouTube, Bilibili, or local video files. Triggers when user asks to summarize, analyze, extract key points from a video, or assess video value. Supports Chinese/English video content analysis with MiniMax LLM.

Video Content Watcher

Extract audio → transcribe → analyze → generate structured Markdown report.

Quick Start

cd /home/ykl/.openclaw/workspace-code-dev/video-reader
export MINIMAX_API_KEY="your-key"
export WHISPER_MODE="local"   # or "api"
python3 << 'EOF'
import sys; sys.path.insert(0, 'src')
from video_reader_mcp import VideoReaderMCP
mcp = VideoReaderMCP()
result = mcp.process_url('VIDEO_URL')
print(result['report'])
EOF

Or use the CLI script:

python3 /home/ykl/.openclaw/workspace/skills/video-content-watcher/scripts/video_reader.py "VIDEO_URL"

Workflow

Audio extraction: Uses yt-dlp for online videos / ffmpeg for local files
Transcription: Whisper (local openai-whisper or API mode)
Analysis: MiniMax LLM analyzes transcript → structured report

Configuration (env vars)

Variable	Required	Default	Description
`MINIMAX_API_KEY`	Yes	—	MiniMax API key
`WHISPER_MODE`	No	`api`	`api` or `local`
`WHISPER_API_KEY`	If mode=api	—	OpenAI API key for Whisper
`LLM_BASE_URL`	No	`https://api.minimaxi.com/v1`	MiniMax endpoint
`LLM_MODEL`	No	`MiniMax-M2.7`	Model name

Supported Sources

YouTube: https://www.youtube.com/watch?v=xxx
Bilibili: https://www.bilibili.com/video/BVxxx
Local files: /path/to/video.mp4

Output

Returns a Markdown report with:

Basic info (source, transcript length)
Core content summary
Key takeaways
Main viewpoints
Pros / cons
Target audience

Scripts

scripts/video_reader.py — CLI entry point
scripts/analyze_text.py — Analyze raw text directly (skip video processing)