whatisgithub

What is chubbyskills?

chubbyguan/chubbyskills — explained in plain English

Analysis updated 2026-05-18

205PythonAudience · developerComplexity · 3/5Setup · moderate

In one sentence

A personal collection of AI agent skills for transcribing videos, managing an Obsidian knowledge base, tracking industry news, and generating flashcards, built around Chinese platforms.

Mindmap

mindmap
  root((chubbyskills))
    What it does
      Video transcription
      Knowledge base upkeep
      News radar
      Flashcard generation
    Tech stack
      Python
      SenseVoice
      faster-whisper
      ffmpeg
    Use cases
      Transcribe Douyin YouTube
      Bilingual transcripts
      Clean Obsidian vault
      Daily keyword briefing
    Audience
      Developers
      Content creators

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Turn a Douyin, Bilibili, or YouTube video link into a Markdown transcript automatically.

USE CASE 2

Get an English YouTube video transcribed and translated into a bilingual Chinese document.

USE CASE 3

Keep an Obsidian vault tidy by auto tagging, deduplicating, and fixing broken links.

USE CASE 4

Track keyword mentions across X, Jike, V2EX, Hacker News, and 36kr in a daily briefing.

What is it built with?

PythonSenseVoice-Smallfaster-whisperffmpeg

How does it compare?

chubbyguan/chubbyskillsnvlabs/mobilitygencvlab-kaist/gld
Stars205207196
LanguagePythonPythonPython
Last pushed2026-02-17
MaintenanceMaintained
Setup difficultymoderatehardhard
Complexity3/54/55/5
Audiencedeveloperresearcherresearcher

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Each skill has its own Python requirements file, and some need system tools like ffmpeg installed separately.

So what is it?

Chubby Skills is a personal collection of AI automation routines, open-sourced by a Chinese content creator who uses them daily. Each routine is packaged as a "skill" that compatible AI agents can load directly, following the AgentSkills.io open standard. The collection works with several AI coding tools including Claude Code, Codex, OpenCode, OpenClaw, and Hermes. The largest category covers video transcription. You give the skill a link to a video from Douyin (the Chinese version of TikTok), Bilibili, TikTok, Weibo, Zhihu, or YouTube, and it downloads the audio, runs it through a speech recognition model called SenseVoice-Small, and saves the result as a Markdown file. The YouTube skill goes one step further: it translates the English transcript into Chinese and outputs a side-by-side bilingual document. There is also a podcast skill that handles RSS feeds from Chinese podcast platforms and uses a different transcription model called faster-whisper. Beyond transcription, the collection includes a WeChat public account skill that fetches article links, converts them to Markdown, and extracts structured observations and follow-up questions. A knowledge base management skill helps maintain an Obsidian note vault: it handles tagging and filing new material, checks for broken links and missing metadata, removes duplicates, and integrates with external tools for search and knowledge graph discovery. Two workflow-level skills round out the set. An industry intelligence radar scans X, Jike, V2EX, Hacker News, and 36kr for user-defined keywords, filters and deduplicates the results, detects emerging trends, and produces a daily briefing. A learning notes automation skill takes transcripts from any of the supported video or podcast sources, pulls out key concepts and facts, generates Anki-compatible flashcards in several formats, and updates a knowledge graph. Installation is handled by a setup script in the repository root. You can install all skills at once or pick individual ones by name. Each skill directory contains its own requirements file for Python dependencies, and some skills need system tools like ffmpeg installed separately.

Copy-paste prompts

Prompt 1
Install the video transcription skill from chubbyskills and transcribe this Bilibili link into Markdown.
Prompt 2
Set up the YouTube skill to produce a bilingual English-Chinese transcript for this video.
Prompt 3
Use the knowledge base skill to clean up and deduplicate my Obsidian vault.
Prompt 4
Run the industry intelligence radar skill to track mentions of a keyword across Chinese and English platforms.
Prompt 5
Generate Anki flashcards from this podcast transcript using the learning notes automation skill.

Frequently asked questions

What is chubbyskills?

A personal collection of AI agent skills for transcribing videos, managing an Obsidian knowledge base, tracking industry news, and generating flashcards, built around Chinese platforms.

What language is chubbyskills written in?

Mainly Python. The stack also includes Python, SenseVoice-Small, faster-whisper.

How hard is chubbyskills to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is chubbyskills for?

Mainly developer.

Open on GitHub → Ask about another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.