What is skill-factory?

Skill Factory is a growing collection of skills for Claude Code, the AI coding assistant. Each skill lives in its own folder containing a SKILL.md file plus supporting scripts, and you install one by copying its folder into a .claude/skills directory. Once installed, you can call a skill with a slash command like /book-deepdive, or just describe what you want in plain language. The repository is organized into categories, and so far it has one: learning, which turns source content into interactive study material. Two skills live here. book-deepdive takes the title and author of a book and produces a folder with a breakdown of the book by parts and topics, including concepts illustrated with inline diagrams, a handful of generated pictures, a searchable glossary, and self-check questions. podcast-deepdive does the same for a YouTube video: give it a link and it produces a breakdown of the conversation, including who is speaking, chapters linked to the exact point in the video, quotes checked against the transcript, notes on where speakers disagreed, a glossary, and questions. Every output page is bilingual, with Russian text and English technical terms and a switch between the two, and each page is a single self-contained HTML file that works without a web server. Both skills work the same way under the hood: several AI agents work in parallel to analyze the source material and check each other's facts, and the results are saved into one JSON file per project. A separate, non-AI Python script then reads that JSON file and builds the final HTML page, so the page can be regenerated at any time without redoing the analysis. To use it, you clone the repository and copy the skill folders you want into your project's .claude/skills directory, or into your home folder to make them available everywhere. Running the tool requires Claude Code with its Workflow feature, the uv Python tool for running schema validators, and, for the podcast skill, yt-dlp to fetch video subtitles. Generating illustrations is optional and needs an OpenRouter API key, without one, the pages are still built, just without the pictures. The project is released under the MIT license.