Video Recap Skills — Turns video into narrated recap using AI, with CapCut/JianYing export, requiring only ffmpeg locally.
Analyzed by Sai Pavan Gopularam · AI · Video Content · View on GitHub
- Stars: 527
- Forks: 97
- Commits last 30 days: 49
- Health: Active (49 commits this month)
- Language: Python
- License: MIT
What It Is
Imagine having an AI assistant that watches your video, understands its content, writes a script, narrates it, and then assembles a complete video recap. That's what Video Recap Skills does. It uses advanced AI models (ASR, VLM, TTS) from Xiaomi MiMo to process videos and generate a structured, narrated output, even allowing export to popular editing software like CapCut or JianYing.
This project solves the problem of time-consuming and skill-intensive video content creation. Instead of manually editing, scripting, and voicing, it automates the entire process, producing production-ready recaps or drafts for further refinement. It's particularly useful for creators needing to quickly summarize long-form content or produce short-form narratives.
License Verdict
MIT License — Build and Sell Freely — Commercial Use Approved • No Copyleft Restrictions
The MIT License allows you to use, modify, and distribute this software for any purpose, including commercial use, without needing to open-source your own code. You can integrate it into proprietary products and sell them.
How to Use It
To get started, you'll need Python 3.10+, ffmpeg installed, and a Xiaomi MiMo API key. Once these prerequisites are met, you can install the skills into your preferred agent environment like Claude Code or Codex CLI with a few simple commands.
Prerequisites:
- Python 3.10+
- ffmpeg (with libass)
- Xiaomi MiMo API Key
Estimated setup time: 10 minutes.
brew install ffmpeg
export MIMO_API_KEY=your-mimo-key
/plugin marketplace add zenstory-ai/video-recap-skills
/plugin install video-recap-skills@video-recapWhat I'd Build With This
AI-Powered Social Media Recaps (micro-saas)
Offer a service where content creators upload their long-form videos (e.g., YouTube lectures, podcasts), and your platform automatically generates short, engaging narrated recaps optimized for TikTok, Instagram Reels, or YouTube Shorts. The output includes a script, voiceover, and an editable CapCut project file, saving creators hours of manual editing. You could focus on specific niches like educational content or gaming.
Effort: 2 Weeks Build Time · Target: Content Creators, Educators · Pricing: $29/mo for 10 videos, $99/mo for unlimited
Automated News & Event Summarizer (saas)
Develop a SaaS platform for media organizations or corporate communications departments to automatically generate narrated video summaries of news broadcasts, internal meetings, or public events. Users upload video footage, and the system produces concise, branded recaps, ready for internal distribution or quick publication. The CapCut export allows for final branding touches by human editors.
Effort: 3 Months Build Time · Target: Media Companies, Corporate Comms · Pricing: $199/mo for basic, custom enterprise plans
E-learning Content Generation Suite (enterprise)
Provide a custom enterprise solution for large e-learning platforms or universities. This suite would integrate with their existing lecture recording systems to automatically transform long lectures into bite-sized, narrated recap modules. It could also translate and re-narrate content for international audiences, leveraging the project's translation capabilities, significantly reducing the cost and time of producing diverse educational materials.
Effort: 6 Months Build Time · Target: Universities, E-learning Platforms · Pricing: Custom Enterprise Contract ($50k+/year)
Sai Pavan Gopularam's Take
This project is a goldmine for anyone looking to automate video content creation, especially for recap-style videos. The structured output and CapCut export are huge time-savers. I'd estimate a basic micro-SaaS generating 100 recaps/month could easily bring in $2,000/month.
Watch Out For
- MiMo API Dependency: This project heavily relies on the Xiaomi MiMo API for ASR, VLM, and TTS. While cost-effective (around 1.3 CNY per video), it's a third-party dependency that requires an API key and incurs usage costs.
- Agent Framework Required: The skills are designed to run within specific agent frameworks like Claude Code, Codex CLI, OpenCode, or OpenClaw. This means you need to be familiar with and set up one of these environments to use the project.
- Primarily Chinese Output: While there's a mention of translating English video to Chinese, the default and primary examples are for Chinese narration and CapCut/JianYing, which are popular in Chinese-speaking markets. Adapting to other languages might require further customization or different TTS/ASR providers.
- FFmpeg with Libass: You need a specific build of FFmpeg that includes libass for burning subtitles. Standard FFmpeg installations might not include this by default, requiring a custom installation or specific package manager options.
I break down trending repos like Video Recap Skills every week — join the newsletter.