phuaky Customization
Before executing, check for user customizations at:
~/.claude/skills/PAI/USER/SKILLCUSTOMIZATIONS/NavalExplain/
If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.
🚨 MANDATORY: Voice Notification (REQUIRED BEFORE ANY ACTION)
You MUST send this notification BEFORE doing anything else when this skill is invoked.
Send voice notification:
curl -s -X POST http://localhost:8888/notify \ -H "Content-Type: application/json" \ -d '{"message": "Running the NavalExplain skill to break down a concept visually"}' \ > /dev/null 2>&1 &Output text notification:
Running the **NavalExplain** skill to break down a concept visually...
This is not optional. Execute this curl command immediately upon skill invocation.
NavalExplain
Interactive visual learning inspired by Naval Ravikant's approach: query multiple AI models in parallel, identify the clearest explanation, then drill down with diagrams, analogies, and visuals until the "aha moment."
Core Principles
- Simplest level first — No jargon. Foundational understanding before complexity.
- User drives the depth — Always ask what doesn't click. Never dump information.
- Visuals are the medium — Mermaid diagrams, ASCII art, analogies. Text walls are failure.
- Multi-model diversity — Different models see different angles. Use that.
Workflow Routing
| Workflow | Trigger | File |
|---|---|---|
| Explain | User pastes text, asks to understand something, or invokes /naval-explain | Workflows/Explain.md |
| DrillDown | User says what doesn't click, asks for more visuals, wants deeper on a concept | Workflows/DrillDown.md |
| Save | User explicitly asks to save, /save, "save this session" | Workflows/Save.md |
Mode Detection
| User Input | Route |
|---|---|
| Pastes a paragraph + "/naval-explain" | → Explain workflow (full multi-model) |
| Pastes text + "help me understand" | → Explain workflow |
| "I don't get the X part" / "what about Y" | → DrillDown workflow |
| "diagram this" / "show me visually" | → DrillDown workflow |
| "draw me an image of this" / "illustrate" | → DrillDown workflow (with Art skill flag) |
| "save this" / "/save" | → Save workflow |
Visual Output Defaults
CRITICAL: NEVER output raw ```mermaid code blocks. They render as unreadable text in the terminal.
| Type | When | Format |
|---|---|---|
| ASCII diagrams | Relationships, flows, processes, hierarchies, comparisons | Monospace box-drawing (primary visual format) |
| Analogies | Abstract concepts that need grounding | Plain text with concrete comparisons |
| Rendered PNG | ONLY when user asks for "render diagram" or "show as image" | Mermaid → PNG via RenderMermaid.sh → display with Read tool |
| AI Images | ONLY when user explicitly asks for illustration/image | Art skill invocation |
ASCII is the default. It renders perfectly in any terminal with zero dependencies. Use box-drawing characters (─ │ ┌ ┐ └ ┘ ├ ┤ ┬ ┴ ┼), arrows (→ ← ↑ ↓ ──▶ ◀──), and tables for all visual explanations.
Examples
Example 1: Understanding a dense paragraph
User: [pastes paragraph about quantum entanglement] /naval-explain
→ Invokes Explain workflow
→ Spawns ClaudeResearcher + GeminiResearcher in parallel
→ Synthesizes: "Here are the 3 key concepts in this paragraph..."
→ Asks: "Which of these doesn't click for you?"
→ User: "I get superposition but not how measurement affects the other particle"
→ Invokes DrillDown workflow
→ Generates Mermaid diagram showing measurement → collapse → correlation
→ Asks: "Does this connection make sense now, or should I try a different angle?"
Example 2: Quick concept breakdown
User: "naval-explain option Greeks in finance"
→ Invokes Explain workflow
→ Multi-model query on option Greeks
→ Returns: "4 key concepts: Delta, Gamma, Theta, Vega..."
→ ASCII art showing how each Greek relates to price movement
→ "Which one feels fuzzy?"
Example 3: Save a session
User: "save this"
→ Invokes Save workflow
→ Saves diagrams + explanations to ~/.claude/MEMORY/LEARNING/{date}_{topic}/
→ Confirms what was saved
Integration
Uses
- Research — ClaudeResearcher + GeminiResearcher for multi-model parallel queries
- Art — AI image generation (ONLY on explicit request)
Feeds Into
- MEMORY/LEARNING/ — Saved sessions become persistent knowledge