learn-nexus（学习联系）

🎯 AI Proactive Triggering (ONBOARDING SKILL)

This is an ONBOARDING skill with MEDIUM PRIORITY (suggest after core onboarding complete).

When to Proactively Suggest (AI MUST check user-config.yaml)

Check learning_tracker.completed.learn_nexus in user-config.yaml. If false:

PROACTIVELY SUGGEST when user:

1. Has completed OTHER onboarding skills (setup-goals, learn-projects, learn-skills) but hasn't done learn-nexus yet - suggest as "graduation" step
2. Asks philosophical questions ("why does Nexus work this way?", "what's the design behind...")
3. Encounters AI patterns (false progress, incomplete reads, over-engineering)
4. Makes common mistakes (creating projects instead of skills, skipping close-session)
5. Expresses interest in mastering the system or best practices
6. After using Nexus for 3+ sessions without completing this skill

Suggestion Pattern (after other onboarding):

💡 You've learned the core concepts! Ready for system mastery? 'learn nexus'
(15 min) covers the philosophy, design principles, common pitfalls, and
expert collaboration techniques.

This is the "graduation" skill - after this, you'll understand Nexus deeply.

Suggestion Pattern (encountering issues):

💡 I notice you're running into [AI pattern / common mistake]. The 'learn nexus'
skill covers exactly how to handle this. Want to take 15 minutes to learn
the system deeply? It'll prevent these issues in the future.

DO NOT suggest if:

• learning_tracker.completed.learn_nexus: true
• User hasn't completed core onboarding yet (suggest core skills first)
• User is new and just getting started (too advanced for first sessions)

Learn Nexus

Master Nexus through understanding its philosophy, design principles, and practical patterns.

Purpose

Help users understand WHY Nexus works the way it does, avoid common mistakes, and collaborate effectively with AI.

Time Estimate: 15-18 minutes

Part A: The Living Knowledge Organism (3 min)

Display:

━━━ CORE PHILOSOPHY ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ALL FILES IN THIS SYSTEM ARE EXECUTABLE — NOT DOCUMENTATION!

Every .md file, .yaml config, and planning document is designed
to be READ, LOADED, and EXECUTED by AI in conversation.

This is not a static knowledge base. It's a living, breathing
organism that guides you through work, adapts to your context,
and evolves with every interaction.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Ask: "Does that click? Every markdown file is essentially code for the AI."

Part B: The 7 Problems Nexus Solves (4 min)

Display one at a time, confirm understanding:

━━━ THE 7 PROBLEMS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

1. AI AMNESIA
   Problem: AI forgets everything every session
   Solution: Memory files auto-load every session (goals.md, etc.)

2. FILE CHAOS
   Problem: AI generates files randomly everywhere
   Solution: 5-folder system with clear boundaries

3. INCONSISTENT RESULTS
   Problem: AI improvises differently every time
   Solution: Skills = saved workflows with exact steps

4. REPEATED WORK
   Problem: Build same workflow multiple times
   Solution: Create skill once, reuse forever

5. CONTEXT OVERLOAD
   Problem: Loading everything slows responses
   Solution: Progressive loading (metadata first, content on-demand)

6. HARD TO LEARN
   Problem: Steep learning curve, overwhelming complexity
   Solution: Optional onboarding skills, smart defaults

7. IMMEDIATE EXECUTION
   Problem: AI builds before understanding requirements
   Solution: Planning mode → separate execution session

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Ask: "Which of these problems have you experienced? That's why Nexus exists."

Part C: The 7 Design Principles (4 min)

Display:

━━━ 7 DESIGN PRINCIPLES ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

1. INSTRUCTION-DRIVEN
   Python script returns COMPLETE instructions.
   AI follows exactly. Zero interpretation.

2. YAML-DRIVEN AUTO-DETECTION
   Everything has metadata describing when to load it.
   AI matches your message → context loads automatically.

3. SKILL-FIRST EXECUTION
   Skills have priority over projects.
   User skills have priority over system skills.

4. PROGRESSIVE DISCLOSURE
   Load minimum at start (metadata only).
   More context just-in-time when needed.

5. STATE IN DATA FILES
   System state tracked in YAML and checkboxes.
   No hidden logic. Transparent. Inspectable.

6. CONTEXT PRESERVATION
   Nothing is lost between sessions.
   close-session saves everything automatically.

7. CONCRETE BEFORE ABSTRACT
   Experience first, explanation after.
   Value delivery before feature teaching.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Ask: "Any principle you want me to explain further?"

Part D: System Pitfalls (2 min)

Pitfall #1: Projects Instead of Skills

WRONG:
📁 weekly-report-week-1/
📁 weekly-report-week-2/
📁 weekly-report-week-3/

RIGHT:
📁 weekly-report/  # ONE skill, used weekly

If creating "name-1", "name-2"... STOP! That's a skill.

Pitfall #2: Skipping Close-Session

Without "done" / "close session":
❌ Progress not saved
❌ Learnings not captured
❌ Next session loses context

Always say "done" when finished.

Pitfall #3: Over-Organizing

WRONG (day 1): 20 nested empty folders
RIGHT (day 1): 3-5 essential folders

Add folders WHEN needed, not before.

Part E: AI Collaboration Patterns (2 min)

Pattern #1: False Progress

AI: "Done! Files created."
(But they don't exist)

Detection:
🔍 "Show me the files"
🔍 "Read that back"

Pattern #2: Incomplete Reads

AI reads 100 of 500 lines, misses critical info.

Detection:
🔍 "Did you read the ENTIRE file?"
🔍 "What's at the end?"

Pattern #3: Over-Engineering

You want simple, AI builds framework.

Detection:
🔍 "Can this be simpler?"
🔍 "What's the minimum needed?"

Part F: Expert Techniques (2 min)

"Yes, And..." Collaboration

You: "Create sales proposal template"
AI: "I'll create it. I notice you target enterprise
     clients - should I add ROI calculator?"

Evaluate additions, don't auto-accept/reject.

Intelligent Friction

You: "Create 10 projects for Q1"
AI: "Before 10 projects: Which unblocks others?
     Should some be skills? Start with 2-3?"

Good friction prevents wasted effort.

The Planning Pattern

Session 1 (Planning - 20-30 min):
- Create project structure
- Fill overview.md, plan.md, steps.md
- NO IMPLEMENTATION

Session 2 (Execution):
- Load planning files (crystal-clear requirements)
- Execute end-to-end
- Minimal intervention needed

Planning optimized for thinking. Execution optimized for doing.

Part G: Mastery Check (2 min)

Quick quiz:

1. Creating "expense-report-jan", "expense-report-feb"... Fix?
2. AI says "Done! All files created." Response?
3. What's the #1 design principle of Nexus?

Answers:

1. ONE "expense-report" skill
2. "Show me the files"
3. Instruction-driven (Python script controls, AI follows)

Finalize

Actions (MUST complete all):

1. Mark skill complete in user-config.yaml: learning_tracker: completed: learn_nexus: true # ADD THIS LINE
2. Display completion: ✅ Learn Nexus Complete! You now understand: • The Living Knowledge Organism philosophy • 7 problems Nexus solves • 7 design principles (instruction-driven, skill-first, etc.) • 3 system pitfalls to avoid • 3 AI patterns to detect • Expert collaboration techniques You're ready to use Nexus at full power. Next steps: • 'create project' - Start a new project • 'create skill' - Capture a workflow • Just work - Nexus guides you
3. Prompt close-session: 💡 When you're done working, say "done" to save progress.

Success Criteria

• User understands the Living Knowledge Organism philosophy
• User knows the 7 problems Nexus solves
• User can name key design principles
• User knows 3 system pitfalls and how to avoid them
• User can detect 3 AI behavior patterns
• User understands expert collaboration patterns
• User passed mastery check quiz
• learning_tracker.completed.learn_nexus: true in user-config.yaml