⚡ EXECUTION INSTRUCTIONS FOR CLAUDE

When this command is invoked, YOU (Claude) must execute these steps immediately: This is NOT documentation - these are COMMANDS to execute right now. Use TodoWrite to track progress through multi-phase workflows.

🚨 EXECUTION WORKFLOW

Phase 1: Execute Sequential Thinking

Action Steps:

1. Determine the requested thinking level from arguments (Light/Medium/Deep/Ultra, default to Light if omitted)
2. Map the level to the totalThoughts budget (Light: 4, Medium: 6, Deep: 8, Ultra: 12)
3. Run mcp__sequential-thinking__sequentialthinking with the user's prompt and chosen budget, logging phases in TodoWrite
4. Deliver a synthesized answer that summarizes reasoning, conclusions, and recommended next steps

📋 REFERENCE DOCUMENTATION

Think Command

Usage: /think [level] [problem/question] or /think [problem/question]

Purpose: Engage in systematic problem-solving using sequential thinking methodology with adjustable computation levels.

Thinking Levels

1. Light (/think light or /think l): 3-4 thoughts - Quick analysis
2. Medium (/think medium or /think m): 5-6 thoughts - Standard reasoning
3. Deep (/think deep or /think d): 7-8 thoughts - Thorough analysis
4. Ultra (/think ultra or /think u): 10+ thoughts - Maximum budget

Default Mode

• Light by Default: /think uses light mode (4 thoughts) for efficient problem-solving
• Level Override: Specify level for different analysis depths

Behavior

Uses the mcp__sequential-thinking__sequentialthinking tool to:

• Break down problems into manageable steps
• Allow for revision and course correction during analysis
• Generate and verify solution hypotheses
• Provide reasoning chains appropriate to the selected level
• Handle multi-step solutions with context preservation

Memory Enhancement: This command automatically searches memory context using Memory MCP for relevant past experiences, solutions, and insights to enhance reasoning quality. See CLAUDE.md Memory Enhancement Protocol for details.

Examples

/think What's wrong with this code?                    # Light: 4 thoughts
/think medium How should I refactor this codebase?     # Medium: 6 thoughts
/think deep What's the root cause of this issue?       # Deep: 8 thoughts
/think ultra Plan architecture for 10M requests        # Ultra: 12+ thoughts

# Short aliases

/think l Quick bug analysis
/think m Design pattern selection
/think d Performance optimization strategy
/think u Complex system architecture

Implementation Notes

• Uses totalThoughts parameter to control thinking depth:
  • Light: totalThoughts: 4
  • Medium: totalThoughts: 6
  • Deep: totalThoughts: 8
  • Ultra: totalThoughts: 12
• Can handle uncertainty and explore alternative approaches
• Supports branching and backtracking in reasoning
• Maintains context across multiple reasoning steps
• Generates concrete, actionable solutions
• Efficient by default with light mode for quick problem-solving