# Teach By Doing

> Never teach a principle without an immediate rep. When designing ANY teaching content — a workshop, a lesson plan, a deck, a course module, a webinar, an explainer video, an educational post — pair every theoretical unit with an embodiment the learner does right now. Use whenever building or reviewing teaching material, lesson plans, workshop flows, course outlines, or training content, OR when the user says 'teach-by-doing', 'add an exercise', 'make it hands-on', 'don't leave it abstract', 'תרגיל לכל עיקרון', 'הטמעה מיידית'.

- Skill: `aviz85/teach-by-doing` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add aviz85/teach-by-doing`
- Raw SKILL.md: https://api.skillmd.com/api/skills/aviz85/teach-by-doing/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: aviz85 (https://skillmd.com/u/aviz85)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/aviz85/teach-by-doing

---


# teach-by-doing — no theory without a rep

The single rule: **every theoretical teaching unit must be paired with an immediate embodiment.**
No principle is "taught" until the learner *did* something with it — on the system, in the
interface, on paper, or in their own body and emotion. Abstract talk evaporates. A rep sticks.

This is a **design-time** skill: it runs while you build teaching material, not while you
deliver it. Its job is to make sure no principle ships naked.

## Why (the research, in one breath)

- **Kolb's experiential cycle** closes only at *active experimentation* — and when you can't
  test for real yet, you "make a plan for when and how." That fallback is the whole point.
  Experiential methods lift retention up to ~75% vs. passive lecture.
- **Generation effect:** a learner who *generates* their own example remembers far more than
  one handed a finished example — and generation pulls in personal experience and emotion as
  retrieval cues. (So a writing/own-example rep beats a slide.)
- **Embodied cognition:** a motor action in the interface strengthens the learning signal and
  primes recall. (So "open the tool and do step 1" beats "here's how it would work.")

Three different sciences, one instruction: **make them do it now.**

## The ladder — pick the highest rung the moment allows

For each teaching unit, climb to the **highest feasible rung**. Never drop below rung 4.

| Rung | Use when | The rep | Anchor |
|------|----------|---------|--------|
| **1 · Real embodiment** | the action is doable *today* on the real system | Do it now on the live agent / computer / account. Real artifact, real feedback. | Kolb — concrete experience |
| **2 · Interface rehearsal** | the real action is relevant weeks ahead (a slow process) | Open the interface and do a *partial* move — navigate, scaffold step 1, dry-run it. | Embodied cognition |
| **3 · Writing rep** | can't touch the system at all | On paper / in a doc: generate your *own* example, fill a blank, write a commitment. | Generation effect |
| **4 · Activating rep** | even writing doesn't fit | A felt decision, a visualization, a stated intention — move an emotional process. | Emotion as retrieval cue |

**Rung 1 is the goal. Rungs 2–4 exist so a "you can't do this yet" principle is never an
excuse to stay abstract.** A future-tense process still gets a present-tense rep.

## What this skill does when you invoke it

You point it at teaching content (a plan, a deck, an outline, a draft). It:

1. **Segments** the content into discrete teaching units — one principle / one small step each.
2. **Classifies** each unit: is it implementable on the system *today*? → assign the highest
   feasible rung.
3. **Generates** a concrete rep for each — 2–5 minutes, specific, doable. Not "reflect on X"
   but "open the tool and ask it Y" / "write your own Z in the notebook."
4. **Flags** any unit with no rep as **incomplete** — a gate the material does not pass. If a
   principle resists *every* rung, that is a signal the principle is too abstract to teach yet;
   sharpen it or cut it.

**Output:** a table — `principle → rung → rep → mental muscle` — ready to drop into the
lesson plan. (See `references/EXAMPLES.md` for the canonical worked example.)

## How to write a good rep

- **Present tense, imperative, specific.** "Ask your agent: *what do you know about my
  business, and what's missing?*" — not "consider what the agent knows."
- **Make it generate, not receive.** The learner produces their *own* example, in their *own*
  context. That is where the memory and the emotion live.
- **One muscle per rep.** Name the muscle ("rerouting everything to one source", "consulting
  instead of commanding"). The name is half the learning.
- **Keep it cheap.** 2–5 minutes. A rep they skip because it's heavy teaches nothing.
- **Sequence the reps into an arc**, so the set tells a story (tour → redirect → furnish →
  consult → make-it-mine), not a checklist of unrelated drills.
- **Prefer doing over watching.** A live demo *by the teacher* is rung 0 — it does not count.
  The learner's own hands are the minimum.

## The test before any teaching material ships

> Walk every principle in the material. For each, can you point to the rep the learner does
> *in the moment*? If even one principle is just words — it hasn't been taught yet.