lesson-generator
Generate and revise curricula, arc syllabi, and Jupyter notebook lessons at any zoom level.
Skill by Peleke
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/peleke-lesson-generator
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: design, data, writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/lesson-generator
What's inside
- Skill name
lesson-generator- Description
- Generate and revise curricula, arc syllabi, and Jupyter notebook lessons at any zoom level. Triggers on "design a curriculum", "I want to learn", "create a course", "generate arc/module syllabus", "create lesson/notebook", "update/revise [artifact]", or when working with educational content. Handles the full pipeline from elicitation interview through course map, arc syllabi, and individual module notebooks — plus targeted revisions with cascade awareness.
- License
- MIT
- Instructions
- 1,564 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- body: 1564 lines (spec recommends under 500)
Files (7)
- README.md
- SKILL.md
- engagement-patterns.md
- module-0.2-outline.md · examples/
- module-0.3-outline.md · examples/
- interactive-templates.md
- video-index.md
SKILL.md (first lines)
# Lesson Generator
A recursive curriculum generation and revision system that operates at three zoom levels:
```
Course (elicitation → curriculum brief → arc map)
↓
Arc (theme + context → module breakdown → publication checkpoints)
↓
Module Lesson (spec → belt system → executable notebook)
+ Revision Mode (update any artifact with cascade awareness)
```
---
## Fractal Concept Loop (Core Structural Principle)
**Every concept, at every zoom level, runs the same four-stage loop: DO → NOTICE → CODE → NAME.**
This is not a narrative suggestion — it's a structural requirement. The loop applies recursively: courses are made of arcs, arcs of modules, modules of sections, sections of **concept primitives** — and every concept primitive gets the full cycle.
### The Loop
| Stage | What Happens | Modality |
|-------|-------------|----------|
| **DO** | Experience it. Hear it, see it, interact with it. No labels yet. See `scaffold-pass` for the **designed failure variant**. | All available senses |
| **NOTICE** | Guided observation. "What did you notice?" Pattern recognition prompts. | Reflective |
| **CODE** | Build/modify/express. Change parameters, break it, make something with it. | Tactile/code |
| **NAME** | Now formalize. Give it its proper name, connect to theory, define it. | Symbolic |
### Concept Primitives
A **concept primitive** is the smallest unit of understanding where confusion is possible. Not "what has a name" but "what could go wrong in someone's head."
**Decomposition heuristic**: Keep asking "could someone be confused about a piece of this?" until the answer is no. That's your base case.
**Examples:**
- "A whole step skips one key" — primitive (you could think it means skip two)
- "E→F has no black key between them" — primitive (you could assume all white keys are equally spaced)
- "The W-W-H-W-W-W-H pattern defines major" — NOT primitive. Decompose into: what W means, what H means, why that specific sequence, what happens if you change one
- "Multiplying by a weight makes something count more" — primitive
- "Weights should sum to 1" — primitive (you could think any weights work)
- "A weighted average" — compound, built from the two primitives above
### Decomposition in Outlines
Before planning the loop for a section, **decompose** it into confusion-primitives:
```
Section: Major Scales
Decomposition:
- "notes have distances between them" (foundational, may already be earned)
- "some distances are bigger than others" (half vs whole)
- "a specific distance pattern makes it sound 'major'" (the sequence)
- "the same pattern starting on any note gives you a new major scale" (transposition)
Each primitive → DO → NOTICE → CODE → NAME
Each primitive → sensory projection (which modalities?)
Primitives chain: each one builds on the previous
```
**Skip mechanism**: If a primitive is already earned from a prior module, reference it and move on. No redundant loops. Track earned primitives across modules the same way jargon-earning is tracked.
### Sensory Projection Matrix
Every concept primitive must be projected into **at least 2 available modalities**. Plan this explicitly in the outline:
| Modality | Music Implementation | Math/ML Implementation |
|----------|---------------------|----------------------|
| **Visual** | SVG diagrams, p5.js waveforms, circle/keyboard viz, animations | Plots, geometric diagrams, FuncAnimation, Plotly 3D |
| **Auditory** | Tone.js synthesis, TidalCycles patterns, Strudel embeds, audio playback | Data sonification, audio feedback on convergence, "sound of a distribution" |
| **Tactile/Code** | Live coding, keyboard interaction, sliders, parameter knobs | ipywidgets, parameter tuning, "build a toy" exercises, code cells |
**Per-primitive planning format:**
```
Primitive: "some distances are bigger than others"
DO: hear C→D then hear E→F (auditory). see gaps on keyboard SVG (visual)
NOTICE: "one jump skips a key, the other doesn't. what's different?"
CODE: slider for step size in semitones, hear + see the result (tactile + auditory + visual)
NAME: "whole step = 2 semitones. half step = 1 semitone."
Modalities: auditory ✓ visual ✓ tactile ✓
```
### Fractal Recursion Across Zoom Levels
The loop applies at every level of the curriculum:
| Level | Unit | Loop Application |
|-------|------|-----------------|
| **Course** | Arc | Each arc has a DO (taste demo), NOTICE (patterns), CODE (build), NAME (formalize) |
| **Arc** | Module | Each module motivates → implements → verifies → names |
| **Module** | Section | Each section opens with experience, ends with formalization |
| **Section** | Concept primitive | Each primitive gets the full DO → NOTICE → CODE → NAME cycle |
The section level is where the loop is **mandatory and explicit**. Higher levels emerge naturally from well-structured sections.
---
## Outline-First Workflow (Module Generation)
**Critical**: Don't generate notebooks directly from arc syllabi. The arc syllabus is too high-level. Instead, use an **outline-first workflow**:
```
Arc syllabus (high-level spec)
↓
Riff with user (war stories, scope, conceptual ordering)
↓
Detailed narrative outline (ready-to-generate artifact)
↓
User approves/tweaks outline
↓
Generate notebook from outline
```
### What Makes an Outline "Ready to Generate"
A ready outline has:From Peleke/brunnr, MIT.
More from Peleke/brunnr
- 0.497score
dossier-generator
Peleke
Turn a built system plus a client/deal context into three linked artifacts: a dual-format dossier (human executive summary + agent-facing implementation…
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
agent-pack
Peleke
Package a piece of completed work into an agent-consumable pack: an entrypoint README, the source documents, and an input-output contract, ready to hand to an…
SkillCFreshness: C★ 1Claude Code +9 - 0.437score
hunter-log
Peleke
Persistence layer for the hunter product discovery pipeline.
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
community-pitch
Peleke
Community pitch engine -- takes persona extraction output and signal scan data, then produces a complete paid community plan with concept, platform…
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
content-planner
Peleke
Scan GitHub activity and buildlog entries, synthesize engineering work into publishable content opportunities with fan-out analysis, and output an…
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
curriculum-planner
Peleke
Interactive elicitation protocol for designing tutorial series.
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
decision-log
Peleke
Structured product decision engine -- takes signal scan output and operator preferences, applies the Signal-to-Decision Pipeline (SDP) framework, and produces…
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
issue
Peleke
Draft well-structured GitHub issues with clear titles, structured descriptions, acceptance criteria, and appropriate labels.
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
chapter-generator
Peleke
Generate a single tutorial chapter from a structured spec.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
linwheel-content-engine
Peleke
Orchestrate LinkedIn content creation through LinWheel's MCP server.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
ax-interview
Peleke
Run a structured AX Interview on an agent-tool interaction session.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
ax-rubric
Peleke
Score agent-facing tool descriptions against the AX Description Rubric.
SkillDFreshness: D★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 14
- Last commit (this folder)
- Mar 14, 2026 (7 months ago)
- Last release
- Mar 12, 2026 (7 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Peleke/brunnr
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Peleke
- 0.497score
dossier-generator
Peleke
Turn a built system plus a client/deal context into three linked artifacts: a dual-format dossier (human executive summary + agent-facing implementation…
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
agent-pack
Peleke
Package a piece of completed work into an agent-consumable pack: an entrypoint README, the source documents, and an input-output contract, ready to hand to an…
SkillCFreshness: C★ 1Claude Code +9 - 0.437score
curriculum-planner
Peleke
Interactive elicitation protocol for designing tutorial series.
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
chapter-generator
Peleke
Generate a single tutorial chapter from a structured spec.
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
community-pitch
Peleke
Community pitch engine -- takes persona extraction output and signal scan data, then produces a complete paid community plan with concept, platform…
SkillDFreshness: D★ 1Claude Code +9 - 0.437score
content-planner
Peleke
Scan GitHub activity and buildlog entries, synthesize engineering work into publishable content opportunities with fan-out analysis, and output an…
SkillDFreshness: D★ 1Claude Code +9
Cite this page
lesson-generator skill. Agent Mart. https://agentmart.dev/skills/peleke-lesson-generator (last updated 2026-10-24)