video-summary_skill

This skill generates concise, structured summaries from videos across platforms by auto-transcribing, extracting insights, and delivering ready-to-use outputs.

Python

2.5k

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill video-summary

_meta.json636 B
SKILL.md16.5 KB

Overview

This skill provides AI-powered video summarization for Bilibili, Xiaohongshu, Douyin, YouTube, and local files. It extracts subtitles or transcribes audio, then generates structured summaries with timestamps, key insights, and actionable takeaways in multiple formats.

How this skill works

Given a video URL or local file, the skill auto-detects the platform, fetches available subtitles via platform-specific methods or uses Whisper for transcription, and chunks long content. It sends processed text to a configured LLM to produce structured outputs (plain text, JSON, or Markdown) and cleans up temporary files automatically.

When to use it

You need a concise, timestamped summary of long videos for quick review
Preparing meeting notes, study aids, or content research from multi-platform videos
Extracting transcripts for search, indexing, or downstream NLP tasks
Generating chapter-by-chapter breakdowns for long-form content
Batch-processing video links programmatically with JSON output

Best practices

Provide a direct video URL; use cookies for gated or region-locked content
Choose Whisper model according to speed/accuracy needs (base recommended)
Use --chapter to enable chapter segmentation for long videos
Configure an LLM provider and API key before first run for immediate summaries
Save custom prompt in ~/.video-summary/prompt.txt to control summary style

Example use cases

Researcher summarizes multiple YouTube lectures into key findings and timestamps
Content creator extracts highlights and action items from competitor videos
Translator or editor exports full transcripts for subtitle creation
Product manager creates briefings from Bilibili tutorials for the team
Automation pipeline converts video content into JSON for indexing and search

FAQ

YouTube, Bilibili, Xiaohongshu, Douyin, and local video/audio files are supported; some short-form platforms rely on transcription rather than native subtitles.

What if no subtitles are available?

Use the --transcribe flag or configure Whisper; choose a model (base for balance) to transcribe audio before summarization.

How do I handle private or region-locked videos?

Provide a cookies file via --cookies or set VIDEO_SUMMARY_COOKIES to a cookies.txt exported from your browser to allow authenticated downloads.

Can I get machine-readable output for automation?

Yes. Use --json to receive structured JSON with title, duration, keyPoints, chapters, and full transcript for programmatic use.