Commands Overview

Claude Commands Library provides 11 slash commands organized into five categories.

Version 4.6 - Anti-Hallucination Contract

• HARD-GATE blocks in all skills — pre-flight checklist before any output
• NEVER rules — explicit per-domain prohibitions (no invented paths, versions, or facts)
• Chain-of-Thought REASONING block required in every Step 0.1 output
• /deploy and /new-stack — universal config from personal-profile.md
• Session commands removed in v4.5 — replaced by Claude Code's built-in auto-memory
• View Changelog →

Quick Reference

Command	Purpose	Speed	Version
/prompt	Basic prompt perfection	~2s	v4.0
/prompt-hybrid	Intelligent with agents	2-30s	v4.0
/prompt-technical	Implementation analysis	5-30s	v4.0
/prompt-research	Multi-agent research	60-180s	v4.0
/prompt-dotnet	.NET project-aware perfection	~3s	v2.0
/prompt-react	React project-aware perfection	~3s	v2.0
/prompt-article	Article writing	Interactive	v2.0
/prompt-article-readme	README generation	~30s	v2.0
/deploy	Project-aware deployment	Interactive	v2.0
/new-stack	Docker stack scaffold	~5s	v2.0
/reflect	Improve skills	5-15s	v2.0

Categories

Prompt Engineering

Commands for transforming ideas into precise, executable prompts.

/prompt - Fast prompt perfection

/prompt Fix the login bug showing null error

Best for: Quick fixes, simple questions, single-file changes.

/prompt-hybrid - Intelligent prompt perfection

/prompt-hybrid Add authentication following existing patterns

Best for: Complex tasks, multi-file changes, pattern detection.

/prompt-technical - Technical implementation

/prompt-technical Implement Redis caching for user queries

Best for: New features, refactoring, architecture decisions.

/prompt-research - Deep analysis

/prompt-research Perform security audit of payment system

Best for: Security audits, performance analysis, architecture review.

/prompt-dotnet - .NET project-aware prompt perfection

/prompt-dotnet Add JWT refresh token endpoint

Scans .csproj, Program.cs, appsettings.json — pre-fills framework, auth, ORM, Docker. Best for: C# APIs, EF Core migrations, middleware, .NET features.

/prompt-react - React project-aware prompt perfection

/prompt-react Add a paginated item list with TanStack Query

Scans package.json, vite.config, tsconfig — pre-fills React version, router, state, base path. Best for: Components, hooks, routing, state management, Vite config.

Content Creation

Commands for generating written content.

/prompt-article - Article wizard

/prompt-article Write about CI/CD best practices

Interactive wizard for blog posts, Medium articles, LinkedIn content.

/prompt-article-readme - README generator

/prompt-article-readme

Analyzes project and generates comprehensive README.

Deployment & Infrastructure

Commands for deploying and scaffolding Docker stacks.

/deploy - Project-aware deployment

/deploy
/deploy my-app

Scans project, reads server config from personal-profile.md, generates complete deploy command sequence.

/new-stack - Docker stack scaffold

/new-stack
/new-stack my-app

Generates docker-compose.yml, nginx.conf, Dockerfile.api, .env, deploy.sh for a new stack.

Utility

Commands for system improvement.

/reflect - Skill improvement

/reflect prompt-technical

Analyzes session and proposes improvements based on corrections and preferences.

How Commands Work

Phase 0: Every Command Starts Here

All prompt commands run Phase 0 validation (AI Fluency aligned):

Step 0.1: Initial Analysis
- Detect language (English, Slovak, etc.)
- Identify prompt type (Task, Question, Bug Fix, etc.)
- Extract core intent

Step 0.11: Quick Delegation Check
- Task Appropriateness: Is this suitable for AI?
- AI Capability Match: Does this match AI strengths?
- Responsibility Awareness: User remains responsible

Step 0.12: Interaction Mode Detection
- Automation Mode: "Fix X", "Add Y"
- Augmentation Mode: "Help me understand", "What's best"
- Agency Mode: "Research X", "Explore"

Step 0.2: Completeness Check (9 criteria)
- Recall known facts from project profile (v4.2)
- Product: Goal, Context (pre-filled), Scope, Requirements, Constraints (pre-filled), Expected Result
- Process: Approach methodology
- Performance: Interaction style, Communication tone

Step 0.3: Clarification Questions
- Skip questions answered by project profile
- Ask only genuinely unknown information
- Present options where needed
- Prioritize by criticality

Step 0.4-0.5: Perfect the Prompt
- Structure into clear format
- Add detected context
- Include constraints

Step 0.6: Approval Gate
- Show perfected prompt
- Diligence Reminder: "You remain responsible..."
- Wait for user confirmation

Step 0.7: Post-Execution Evaluation (NEW v4.2)
- Prompt: "How was this output? good/partial/wrong"
- Feedback loop for iteration

Complexity Detection

Commands with hybrid intelligence score your prompt:

Score	Category	What Happens
0-4	Simple	Fast inline validation only
5-9	Moderate	Claude asks if you want agent help
10-19	Complex	Agent spawns automatically
20+	Research	Multi-agent deep analysis

Agent Types

When complexity warrants, specialized agents help:

• Explore Agent (Haiku, 30s) - Scans codebase for patterns
• Plan Agent (Sonnet, 60s) - Creates implementation strategies
• Security Agent (Sonnet, 45s) - Checks for vulnerabilities
• Performance Agent (Sonnet, 45s) - Identifies bottlenecks
• Pattern Agent (Haiku, 30s) - Ensures consistency

Choosing the Right Command

Decision Flow

┌─────────────────────────────────────────┐
│ What do you need?                       │
└─────────────────────────────────────────┘
          │
    ┌─────┴─────┐
    ▼           ▼
┌───────┐   ┌─────────┐
│ Code  │   │ Content │
└───────┘   └─────────┘
    │           │
    │      ┌────┴────┐
    │      ▼         ▼
    │  ┌───────┐ ┌────────┐
    │  │Article│ │ README │
    │  └───────┘ └────────┘
    │      │         │
    │      ▼         ▼
    │  /prompt   /prompt-
    │  -article  article-
    │            readme
    │
    ├─── Simple fix? ───────► /prompt
    │
    ├─── Need codebase ─────► /prompt-hybrid
    │    context?
    │
    ├─── Implementation ────► /prompt-technical
    │    planning?
    │
    └─── Deep analysis? ────► /prompt-research

By Use Case

Use Case	Command
Fix a bug	/prompt
Add a small feature	/prompt or /prompt-hybrid
Add a complex feature	/prompt-technical
Refactor code	/prompt-technical
Security audit	/prompt-research
Performance optimization	/prompt-research
Architecture review	/prompt-research
Write documentation	/prompt-article
Generate README	/prompt-article-readme
Deploy to server	/deploy
Scaffold Docker stack	/new-stack
Improve a command	/reflect

Next Steps

• Tutorial: Getting Best Results - Complete guide
• Learn /prompt - The foundation
• Learn /prompt-technical - For implementation
• AI Fluency Framework - The 4Ds explained
• Understand the architecture - How it all works