---

name: smart-routing description: > Intelligent request routing for /toh command. Analyzes user intent, assesses confidence, detects IDE environment, and routes to the appropriate agent(s). Memory-first approach ensures context awareness. Triggers: /toh command, natural language requests, ambiguous inputs.

Smart Routing Skill

Intelligent routing engine for the /toh smart command. Routes any natural language request to the right agent(s).

---

🧠 Routing Pipeline

Copy┌─────────────────────────────────────────────────────────────────┐
│                    USER REQUEST                                 │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  STEP 0: MEMORY CHECK (ALWAYS FIRST!)                          │
│  ├── Read .toh/memory/active.md                                │
│  ├── Read .toh/memory/summary.md                               │
│  ├── Read .toh/memory/decisions.md                             │
│  └── Build context understanding                               │
│                                                                 │
│  STEP 1: INTENT CLASSIFICATION                                 │
│  ├── Pattern matching (keywords, phrases)                      │
│  ├── Context inference (from memory)                           │
│  └── Scope detection (simple/complex)                          │
│                                                                 │
│  STEP 2: CONFIDENCE SCORING                                    │
│  ├── HIGH (80%+) → Direct execution                            │
│  ├── MEDIUM (50-80%) → Plan Agent first                        │
│  └── LOW (<50%) → Ask for clarification                        │
│                                                                 │
│  STEP 3: IDE DETECTION                                         │
│  ├── Claude Code → Parallel execution enabled                  │
│  └── Other IDEs → Sequential execution only                    │
│                                                                 │
│  STEP 4: AGENT SELECTION & EXECUTION                           │
│  └── Route to appropriate agent(s)                             │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

---

📊 Intent Classification Matrix

Primary Patterns → Agent Mapping

---

🎯 Confidence Scoring Algorithm

Copyinterface ConfidenceFactors {
  keywordMatch: number;      // 0-40 points
  contextClarity: number;    // 0-30 points
  memorySupport: number;     // 0-20 points
  scopeDefinition: number;   // 0-10 points
}

function calculateConfidence(request: string, memory: Memory): number {
  let score = 0;
  
  // Keyword matching (0-40 points)
  // Strong match with primary patterns = 40
  // Partial match = 20
  // No match = 0
  score += keywordMatchScore(request);
  
  // Context clarity (0-30 points)
  // Specific page/component mentioned = 30
  // General area mentioned = 15
  // No specifics = 0
  score += contextClarityScore(request);
  
  // Memory support (0-20 points)
  // Request relates to active task = 20
  // Request relates to project = 10
  // No memory context = 0
  score += memorySupportScore(request, memory);
  
  // Scope definition (0-10 points)
  // Single clear task = 10
  // Multiple related tasks = 5
  // Unclear scope = 0
  score += scopeDefinitionScore(request);
  
  return score; // 0-100
}

// Thresholds
const HIGH_CONFIDENCE = 80;    // Execute directly
const MEDIUM_CONFIDENCE = 50;  // Route to Plan Agent
// Below 50 = Ask for clarification

---

🖥️ IDE Detection

Detection Method

Copyfunction detectIDE(): 'claude-code' | 'cursor' | 'gemini' | 'codex' | 'unknown' {
  // Check for IDE-specific markers
  
  // Claude Code detection
  if (hasClaudeCodeMarkers()) {
    return 'claude-code';
  }
  
  // Cursor detection
  if (hasCursorRules()) {
    return 'cursor';
  }
  
  // Gemini CLI detection
  if (hasGeminiConfig()) {
    return 'gemini';
  }
  
  // Codex CLI detection
  if (hasCodexConfig()) {
    return 'codex';
  }
  
  return 'unknown';
}

Execution Strategy by IDE

---

🔄 Routing Decision Tree

CopyRequest arrives
      │
      ▼
┌─────────────────────────────────────┐
│ 1. Load Memory Context              │
└─────────────────────────────────────┘
      │
      ▼
┌─────────────────────────────────────┐
│ 2. Is request "continue"/"ทำต่อ"?   │
├── YES → Read memory, resume task   │
└── NO → Continue analysis           │
      │
      ▼
┌─────────────────────────────────────┐
│ 3. Calculate Confidence Score       │
└─────────────────────────────────────┘
      │
      ├── Score >= 80 (HIGH)
      │   └─→ Select agent based on intent
      │       └─→ Execute directly
      │
      ├── Score 50-79 (MEDIUM)
      │   └─→ Route to Plan Agent
      │       └─→ Plan Agent analyzes & routes
      │
      └── Score < 50 (LOW)
          └─→ Ask clarifying question
              └─→ Wait for user response

---

📋 Clarification Patterns

When to Ask

When NOT to Ask

---

🎨 Skill Loading by Intent

Note: response-format skill is ALWAYS loaded for proper output formatting.

---

💾 Memory Integration

Pre-Routing Memory Check

CopyBefore routing, ALWAYS:
1. Read .toh/memory/active.md
   - Current task context
   - In-progress work
   - Blockers
   
2. Read .toh/memory/summary.md
   - Project overview
   - Completed features
   - Tech stack used
   
3. Read .toh/memory/decisions.md
   - Past architectural decisions
   - Design choices
   - Naming conventions

Use memory to:
- Boost confidence (if request matches active work)
- Provide context (for ambiguous "it" references)
- Maintain consistency (follow established patterns)

Post-Execution Memory Save

CopyAfter routing completes, ALWAYS:
1. Update .toh/memory/active.md
   - Mark completed items
   - Update current focus
   - Set next steps
   
2. Add to .toh/memory/decisions.md
   - If new decisions were made
   
3. Update .toh/memory/summary.md
   - If feature was completed

⚠️ NEVER finish without saving memory!

---

📌 Examples

Example 1: High Confidence → Direct

CopyRequest: "/toh สร้างหน้า dashboard"

Analysis:
- Keyword match: "สร้าง" + "หน้า" = Create UI (40 pts)
- Context clarity: "dashboard" = specific page (30 pts)
- Memory: Project has other pages (15 pts)
- Scope: Single page (10 pts)
Total: 95 pts = HIGH

Route: UI Agent (direct)

Example 2: Medium Confidence → Plan First

CopyRequest: "/toh build e-commerce"

Analysis:
- Keyword match: "build" = Create (40 pts)
- Context clarity: "e-commerce" = general concept (10 pts)
- Memory: New project (0 pts)
- Scope: Multiple features (0 pts)
Total: 50 pts = MEDIUM

Route: Plan Agent first → then execute plan

Example 3: Low Confidence → Ask

CopyRequest: "/toh fix it"

Analysis:
- Keyword match: "fix" (20 pts)
- Context clarity: "it" = unclear (0 pts)
- Memory: No recent bugs (0 pts)
- Scope: Unknown (0 pts)
Total: 20 pts = LOW

Action: Ask "What would you like me to fix? Please describe the issue."

---

⚠️ Critical Rules

1. Memory ALWAYS first - Never route without checking context
2. Confidence drives action - Trust the scoring system
3. Plan Agent is your friend - When in doubt, route to Plan
4. IDE awareness matters - Parallel only in Claude Code
5. response-format always loaded - Every response needs 3 sections

---

Smart Routing Skill v1.0.0 - Intelligent Request Routing Engine