An interactive, turn-by-turn guided learning tutor skill for AI coding agents. Teaches any topic in small chunks with visual aids, builds a lesson plan, and checks understanding with edge-case questions — one turn at a time.
🚧 Work in Progress — The core skill works, but more reference transcripts and refinements are coming.
npx skills add static-var/guided-learningWhen a user asks to learn something ("Teach me X", "Explain Y", "How does Z work?"), the agent enters a structured teaching loop:
- Builds a Lesson TODO — 5–12 actionable items, updated every turn
- Teaches one chunk — 2–3 short paragraphs, no walls of text
- Adds a visual — table, mermaid diagram, or ASCII art (one per turn, only when helpful)
- Asks one question — edge cases, predictions, comparisons to force thinking
- Stops and waits — grades the user's answer next turn, then continues
The skill handles continuations (answers, "next", hints, branch picks) and resets cleanly on topic changes.
README.md ← you are here (not installed to agents)
LICENSE ← MIT (not installed to agents)
skills/
guided-learning/
SKILL.md ← installed to agents
references/
style-examples.md
transcripts/
01-kotlin-coroutines.md
02-kotlin-flow.md
Only the skills/guided-learning/ directory is installed into agent skill directories. The root README and LICENSE stay on GitHub.
Works with any agent that supports the Agent Skills format, including:
Claude Code, GitHub Copilot, Cursor, Windsurf, Codex, Cline, Roo Code, Gemini CLI, Trae, and many more.
MIT