Design, refine, and validate technical tutorials as runnable learning experiments. Use when creating or improving course notebooks, technical lessons, demo code, exercises, teaching cases, or student-facing documentation across any technology stack.
インストール
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Design, refine, and validate technical tutorials as runnable learning experiments. Use when creating or improving course notebooks, technical lessons, demo code, exercises, teaching cases, or student-facing documentation across any technology stack.
Technical Tutorial Designer
Use this skill to turn a technical topic into a student-facing, runnable tutorial. The tutorial must build a concept, carry it with a realistic case, and prove it with executable evidence.
Core Principle
A technical tutorial is not a knowledge-point article. It is a runnable learning experiment:
Build the concept.
Create the need through a case.
Implement the concept in code.
Prove the result with observable output.
Required Tutorial Structure
Every tutorial should follow this order unless the user explicitly asks for a different format:
Core Concept: One-sentence definition, problem type it solves, and where it sits in the technical system.
Concept Boundary: What it is not, common confusions, and when not to use it.
Case Background: Realistic business or engineering scenario, user questions, and expected workflow.
Design Thinking: Ask the learner to predict modules, data flow, state, tools, routes, errors, or tradeoffs before implementation.
Data / Environment Setup: Local reproducible data, dependencies, desensitized samples, and scale.
Naive / Bad Example: A natural but flawed implementation that exposes the pain point.
Problem Observation: Use output, numbers, errors, performance, state changes, or complexity to show the issue.
Core Implementation: Implement the target concept step by step. Code must stay tied to the case.
Runnable Chain: Execute the full path, not just define functions or classes.
Comparative Output: Show before/after differences with concrete evidence.
Final Demo + Exercises: One final cell/block summarizes all key capabilities. Exercises only extend the main concept.
Writing Rules
Write for students, not for the author or instructor.
Do not include author-facing notes in the tutorial body. See references/checklist.md for the blocked wording list.
Do not use toy data when the concept depends on scale, privacy, state, latency, errors, or integration.
Do not define APIs/tools/classes without running them in a visible chain.
Prefer one strong concept per lesson. Supporting techniques must serve the main concept.
The final demo should read like an acceptance report: it must show what worked and the evidence.
Validation Workflow
Before final handoff:
Run the tutorial from a clean state when feasible.
