Introduction

Claude Commands Library is a collection of slash commands for Claude Code that transform how you write prompts. Instead of hoping Claude understands your intent, every prompt goes through a validation process that ensures clarity before execution.

Why Use This Library?

The Traditional Approach

You: "Add authentication to my app"
Claude: *implements basic username/password*
You: "No, I meant OAuth with Google"
Claude: *refactors to OAuth*
You: "And I need JWT tokens for the API"
Claude: *adds JWT layer*

Result: 3 rounds of back-and-forth, wasted tokens, fragmented implementation.

The Claude Commands Approach

You: /prompt-technical Add authentication to my app

Claude: Before I implement, let me clarify:

1. Which authentication method?
   - Option A: OAuth (Google, GitHub, Microsoft)
   - Option B: JWT with username/password
   - Option C: Session-based with cookies
   Recommended: Option A for user convenience

2. Which providers should I support?
   - Google (most common)
   - GitHub (developer audience)
   - Microsoft (enterprise)

3. Where should I store user data?
   - Your existing Users table
   - New OAuth-specific tables

You: A, Google only, existing Users table

Claude: *implements exactly what you need, first time*

Result: One clear conversation, precise implementation.

Core Concepts

Phase 0: Prompt Perfection

Every command begins with Phase 0 - a validation process that:

1. Detects what you're asking (task, question, bug fix, etc.)
2. Checks for missing information using 6 criteria
3. Asks clarifying questions before proceeding
4. Structures your request into an unambiguous format
5. Waits for your approval

Learn more about Phase 0 →

Complexity Detection

Not every task needs deep analysis. The library automatically scores complexity:

Score	Category	Action
0-4	Simple	Fast inline validation
5-9	Moderate	Asks if you want agent help
10-19	Complex	Spawns Explore Agent automatically
20+	Research	Multi-agent deep analysis

Agent Types

When complexity warrants it, specialized agents help:

• Explore Agent - Scans your codebase for patterns and conventions
• Plan Agent - Creates implementation strategies
• Security Agent - Checks for OWASP vulnerabilities
• Performance Agent - Identifies bottlenecks
• Pattern Agent - Ensures consistency with existing code

What You'll Learn

This documentation covers:

1. Installation - Get set up in 2 minutes
2. Quick Start - Your first commands
3. Your First Prompt - Detailed walkthrough
4. Commands - Complete command reference
5. Architecture - How everything works
6. Best Practices - Get the most from the library

Requirements

• Claude Code CLI installed
• Windows 10/11 or compatible PowerShell environment
• Git (for installation)

Next Steps

Ready to get started?

Recommended Path

1. Install the library (2 minutes)
2. Try the quick start (5 minutes)
3. Build your first perfect prompt (10 minutes)

Begin Installation →