Where every command is a performance, and every performance delivers results.

Welcome to a curated collection of Claude Code command templates — your personal repertoire of battle-tested prompts that turn Claude into a specialized coding assistant for any task. Think of these as your greatest hits collection, but for getting stuff done with AI.

Each command in this theater follows the "Make It Work First" philosophy outlined in our MANIFESTO.md — because the best code is code that actually works, not code that's drowning in defensive paranoia.

• think.md - The Strategist
  Creates focused step-by-step plans and todo lists. Perfect for when you need to break down complex requests into manageable chunks.

• do.md - The Project Manager
  Executes tasks with multiple specialized agents working in parallel. Your go-to for getting things done with proper orchestration.

• save.md - The Archivist
  Saves conversation context to persistent memory using claude-mem tools. Never lose important insights again.

• remember.md - The Detective
  Searches claude-mem to find relevant context and previous work. Your AI has amnesia, but this gives it perfect recall.

• logic-blocks.md - The Logic Block Detective
  Self-contained logic block review system focusing on control flow, decision trees, and cognitive complexity. Works independently to analyze code logic structures and integrates seamlessly with the logic-flow-auditor agent.

• design-review-technical.md - The System Architect
  Comprehensive technical design reviews focusing on root causes of visual inconsistencies and architectural issues.

• design-review-visual.md - The Visual Detective
  Identifies visual inconsistencies and design system gaps with systematic analysis.

This collection is built on three core principles:

Stop building defensive fortresses. Write code that does the thing, then fix real problems when they actually happen.

• Keep It Simple, Stupid - Complexity is the enemy
• Don't Repeat Yourself - But don't abstract prematurely either
• You Ain't Gonna Need It - Build what you need when you need it

These commands work with specialized agents:

• 🧠� general-purpose - Your primary workhorse
• 📄 docs-alignment-enforcer - Keeps code following best practices
• 🌵 dry-architecture-auditor - Eliminates code duplication
• 💭 logic-flow-auditor - Spots over-complex patterns
• 😵‍💫 confusion-eliminator - Fixes architectural chaos
• 🪄 claude-md-wizard - Masters CLAUDE.md files

1. Pick your command based on what you want to accomplish
2. Replace #$ARGUMENTS with your specific requirements
3. Fire it off to Claude Code
4. Watch the magic happen

# Want to plan a new feature?
cat think.md | sed 's/#$ARGUMENTS/implement user authentication system/'

# Need to execute a complex task?
cat do.md | sed 's/#$ARGUMENTS/refactor the entire API layer to use TypeScript/'

# Time for a logic block review?
cat logic-blocks.md  # This one guides you interactively

• Chain commands: Use save.md after complex work, then remember.md in future sessions
• Parallel execution: do.md excels at orchestrating multiple agents simultaneously
• Context matters: Always provide clear, specific arguments — these commands are only as good as your input

Each command follows a consistent pattern:

<!-- TASK -->
[Clear description of what this command does]

<!-- RULES -->
[Specific guidelines and constraints]

<!-- CONFIGURATION -->
[Agent specifications and special instructions]

The #$ARGUMENTS placeholder gets replaced with your specific requirements, making each command infinitely customizable while maintaining consistent behavior.

• Built for Claude Code CLI - These templates are optimized for the official Claude Code interface
• Agent-Aware - Designed to work with Claude's specialized agent system
• Memory Integration - Several commands integrate with claude-mem for persistent context
• Philosophy-Driven - Every command embodies the "Make It Work First" mindset

Instead of reinventing the wheel every time you need Claude to do something complex, these templates give you:

• Consistency - Same high-quality results every time
• Efficiency - No more crafting the perfect prompt from scratch
• Specialization - Each command is tuned for specific use cases
• Orchestration - Built-in agent management and parallel execution
• Memory - Persistent context across sessions

Have an idea for a new command template? The pattern is simple:

1. Define the task clearly
2. Specify the rules and constraints
3. Choose the right agents
4. Test with real work
5. Add it to the collection

Don't forget to read the MANIFESTO.md — it's not just philosophy, it's the secret sauce that makes these commands cut through complexity like a hot knife through butter.

Remember: Every line of defensive code you write before proving your feature works is a lie you tell yourself about problems that don't exist.

Make it work first. Make it work always. Make guards earn their keep.