AgentSkillsCN

lessons

引导用户循序渐进地学习AI助手的基础知识。

SKILL.md
--- frontmatter
name: lessons
description: Guide users through progressive lessons to learn AI assistant basics
allowed-tools: ["Read", "Write", "Bash"]

Sourdough Lessons

Progressive lessons that teach users how to work with AI assistants by actually using the system.

Purpose

Enable users to learn AI augmentation through hands-on experience, not just reading documentation. Lessons build on each other and teach core concepts through practical exercises.

When to Use

Automatic invocation:

  • When user says "start lessons", "teach me", "show me lessons", "lesson 1", etc.
  • When user asks "what can I learn?" or "how do I get started?"

Progressive invocation:

  • After completing a lesson, suggest the next lesson
  • Track progress in user's data layer

Available Lessons

Lessons are stored in skills/lessons/content/ and numbered for progressive learning:

  1. Welcome & First Conversation - Basics of talking with AI
  2. Reading Files - How AI can read and understand files
  3. Creating Files - Having AI write files for you
  4. Editing Files - Modifying existing files safely
  5. Understanding Context - How AI remembers and uses context
  6. Running Commands - Using bash/PowerShell through AI
  7. Troubleshooting Together - Collaborative problem-solving
  8. Risk & Responsibility - Understanding file access implications
  9. Skills System - Using and creating skills
  10. Session Logging - Tracking your work over time
  11. Learning Framework - How AI learns your preferences
  12. Working with Projects - Managing multiple efforts
  13. Best Practices - Patterns that work well
  14. Next Steps - Where to go from here

Process

1. Determine Intent

Check what the user wants:

  • Start from beginning? → Lesson 1
  • Specific lesson number? → Load that lesson
  • Continue where they left off? → Check progress file
  • List available lessons? → Show lesson menu

2. Load Progress (if exists)

Check for progress tracking file:

bash
PROGRESS_FILE="${SOURDOUGH_DATA:-$HOME/ai-data}/learning/lesson_progress.yaml"

If file exists, read it to know:

  • Last completed lesson
  • Current lesson in progress
  • Notes from previous lessons

3. Load Lesson Content

Read the appropriate lesson file from skills/lessons/content/NN-name.md

Present the lesson content to the user, following the structure in the lesson file.

4. Interactive Teaching

Lessons are interactive experiences with pause points where you wait for user action:

Structure:

  • Present a section up to a [PAUSE] marker
  • Wait for user to try the exercise or ask questions
  • Respond to what they did with feedback and encouragement
  • Present next section up to next [PAUSE]
  • Continue until lesson complete

During pauses:

  • Don't rush ahead - let them experiment
  • Give specific feedback on what they tried
  • If they get stuck, offer hints or demonstrations
  • If they ask questions, answer them before continuing
  • Celebrate when they succeed ("Nice! That's exactly right.")

Tone:

  • Friendly and conversational
  • Encouraging, not condescending
  • Patient with mistakes
  • Enthusiastic about progress

5. Check Understanding

At the end of each lesson:

  • Ask if user has questions
  • Offer to demonstrate again if needed
  • Suggest they try the concept on their own
  • Confirm they're ready to move forward

6. Update Progress

After lesson completion, update progress file:

yaml
last_updated: YYYY-MM-DD HH:MM
completed_lessons:
  - 01-first-conversation
  - 02-reading-files
current_lesson: 03-creating-files
notes:
  - "User particularly interested in file operations"
  - "Completed lesson 2 quickly, comfortable with reading"

7. Suggest Next Steps

Offer:

  • Continue to next lesson immediately
  • Take a break and come back later
  • Practice current lesson concepts before moving on
  • Skip ahead if they're comfortable (with caution)

Progress Tracking Location

User's progress is stored in their personal data layer:

  • MacOS/Linux: ~/ai-data/learning/lesson_progress.yaml
  • Windows: %USERPROFILE%\ai-data\learning\lesson_progress.yaml

Create the file on first use if it doesn't exist.

Output Format

Lesson Presentation

markdown
# Lesson N: [Title]

[Friendly introduction to the lesson topic]

## What You'll Learn

- Key concept 1
- Key concept 2
- Key concept 3

## Let's Try It

[Interactive demonstration or exercise]

## Key Takeaways

- Important point 1
- Important point 2

## Questions?

[Pause for user questions]

---

Ready for the next lesson? Or would you like to practice this more?

Lesson Menu

markdown
# Sourdough Lessons

Your progress: [X] of [14] lessons completed

## Available Lessons

✅ 1. Welcome & First Conversation (completed)
✅ 2. Reading Files (completed)
→ 3. Creating Files (in progress)
   4. Editing Files
   5. Understanding Context
   ... [etc]

Which lesson would you like to start?

Special Lesson Notes

Lesson 8: Risk & Responsibility

This lesson is critical and should not be skipped. It covers:

  • What file access means in practice
  • What could go wrong
  • User responsibility model
  • How to stay safe
  • When to say "no" to AI suggestions

Present this seriously but not alarmingly. Goal is informed users, not scared users.

Lesson 14: Next Steps

Graduates users from lessons to independent usage. Includes:

  • Recap of key concepts
  • Resources for continued learning
  • Suggestions for customization
  • Community/support information

Notes

  • Keep lessons short (5-10 minutes each)
  • Make them practical, not theoretical
  • Encourage experimentation
  • Celebrate progress
  • Be patient with questions
  • Remember: learning happens by doing

Philosophy

"The best way to learn AI augmentation is to be augmented by AI while learning about AI augmentation."


Pro tip: If a user struggles with a concept, break it down further or demonstrate it live. There's no rush—understanding matters more than speed.