A Python-based collection of Claude Code skills organized by category. Provides learning and podcast-deepdive skills, setup instructions, and validation workflow using JSON schemas.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (2 so far).
What it is
A curated collection of Claude Code skills, organized by category. Each skill folder contains a SKILL.md and scripts, and is copied into .claude/skills for use via /<skill-name> commands or natural phrases.
How it works
Skills are grouped into categories (e.g., learning). Each skill outputs a JSON file (book.json or podcast.json) as the contract, which is validated by a validator using JSON Schema. A deterministic Python script reads these outputs to generate an HTML page. The process uses multi-agent analysis to produce the final content.
Getting started
Installation steps:
git clone https://github.com/mldogs/skill-factory
mkdir -p your-project/.claude/skills
cp -R skill-factory/learning/book-deepdive your-project/.claude/skills/
cp -R skill-factory/learning/podcast-deepdive your-project/.claude/skills/
For global installation across projects, copy into ~/.claude/skills/. After restart, skills are available as /book-deepdive and /podcast-deepdive.
Recent releases
- none
Traction
29 stars, 2 forks, 0 open issues. License MIT. Created 2026-07-15, last push 2026-07-15.
Behind the repo
Not provided
Caveats
License: MIT
Open issues: 0
Dependencies include Claude Code with Workflow, uv for validators, and yt-dlp for podcast-deepdive. An environment file .env at repo root is used for illustration generation with keys like OPEN_ROUTER_API_KEY and optional image model/style settings. Generated images are cached on disk; without a key, skills still work and illustrations are omitted.