Getting Started

New to Mannay Claude Code? This guide explains everything.

What Is This?

A development plugin that makes Claude Code better at building software. Think of it like giving Claude:

• A brain - Systematic processes (skills)
• Expert friends - Domain specialists (agents)
• Power tools - Quick scaffolding (commands)
• Autopilot - Optional autonomous execution (Ralph Mode)
• Memory - Persistent state across sessions (Memory System)

The 3 Layers

Layer 1: Skills = The Process

Skills are systematic workflows that auto-activate.

Example:

• You say “Add dark mode” → brainstorming activates → explores design first
• You say “Fix this bug” → systematic-debugging activates → finds root cause

Layer 2: Agents = The Experts

Agents are domain specialists invoked by skills or directly.

Example:

Building auth feature:
1. brainstorming explores approach
2. security-engineer reviews for vulnerabilities
3. backend-architect designs API
4. TDD implements with tests
5. code-reviewer checks quality

Layer 3: Commands = Quick Tools

Commands are slash commands for quick actions.

/api-new users endpoint     # Scaffold API route
/component-new UserCard     # Create React component
/page-new dashboard         # Create Next.js page
/lint                       # Run linting

See Commands Reference for full list.

Smart Orchestration (v1.3.0)

Mandatory Behaviors

These skills activate automatically - you don’t need to ask:

Multi-Agent Chaining

Tasks use ALL relevant agents, not just one:

"Add user authentication"
  → brainstorming (approach)
  → security-engineer (vulnerabilities)
  → backend-architect (API design)
  → api-designer (contracts)
  → typescript-pro (types)
  → TDD (implementation)
  → code-reviewer (quality)

Intent Detection

Mannay understands what you want, not just keywords:

Ralph Mode (Optional)

For autonomous execution when you want Claude to work independently.

When to Use Ralph

✅ Good for:

• Well-defined PRDs with clear stories
• Overnight/weekend work
• Sequential independent tasks

❌ Not for:

• Exploratory work
• Major architecture decisions
• Security-critical code needing human review

Ralph Workflow

1. /ralph-init          # Initialize Ralph files
2. /ralph-build         # Create PRD interactively
3. /ralph-start         # Start autonomous loop
4. Come back later      # Work is done with commits

Memory System

Mannay includes automatic persistent memory that works in the background.

What It Does

How It Works

1. During work - Tracks files modified and decisions made
2. Before compaction - Saves state as a “handoff”
3. After compaction - Restores state and continues
4. On session end - Extracts learnings for future use

Commands

/memory-status      # Check system status
/memory-learnings   # View accumulated learnings

See Memory System for full documentation.

Next Steps

1. Quick Start - Install in 5 minutes
2. Skills Reference - All skills explained
3. Cheatsheet - Quick reference