😎 Claude Vibes

A Claude Code plugin for people who describe WHAT they want and let Claude handle HOW.

It covers three things that Claude Code does not ship on its own:

• Build a brand from nothing. Sixteen guided commands take you from founder interview to a compiled brand guidelines document: purpose, values, positioning, archetype, voice, messaging, tagline, pitch, colors, and typography.
• Turn an idea into a plan. Discovery, scoping, architecture, and a roadmap, explained in plain language for people who don't read code.
• A creator toolkit. Sponsor scripts, hooks, marketing copy, deep research, and image prompts for Midjourney and Nano Banana Pro.

It also wraps building, shipping, debugging, and refactoring in the same plain-language, confirm-before-acting style, and leans on Claude Code's built-in review, reasoning, and memory wherever those exist.

📦 Installation

Open your project in your IDE of choice (VS Code, Cursor, etc.) and open the integrated terminal.

1. Install Homebrew

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

2. Install Node.js

Node.js 18+ is required for MCP servers.

brew install node

3. Install Claude Code

brew install --cask claude-code

4. Install Claude Vibes

mkdir -p .taskmaster && printf '{\n  "models": {\n    "main": {\n      "provider": "claude-code",\n      "modelId": "opus"\n    }\n  }\n}\n' > .taskmaster/config.json && grep -qxF '.taskmaster/' .gitignore 2>/dev/null || echo '.taskmaster/' >> .gitignore && claude "/plugin marketplace add mike-coulbourn/claude-vibes"

5. Start Using Claude Vibes

Type /exit, then run claude again to load the plugin. Run /help to see your new commands.

Local Install (Alternative)

git clone https://github.com/mike-coulbourn/claude-vibes.git

Add as a local marketplace:

/plugin marketplace add ./claude-vibes

Team Installation

Add to your repository's .claude/settings.json for automatic team setup:

{
  "extraKnownMarketplaces": {
    "claude-vibes": {
      "source": {
        "source": "github",
        "repo": "mike-coulbourn/claude-vibes"
      }
    }
  },
  "enabledPlugins": {
    "claude-vibes@claude-vibes": true
  }
}

When team members trust the repository folder, the plugin installs automatically.

🔄 The Workflow

BRAND → START → BUILD → SHIP → FIX → REFACTOR        + TOOLKIT (any time)
  ↓       ↓       ↓       ↓      ↓        ↓
Identity Plan   Code   Deploy  Debug   Evolve

Each phase has dedicated commands and agents designed for that stage of development.

🤖 Commands & Agents

🎨 00-BRAND (Brand Identity)

Create complete brand identities for startups. A structured workflow from discovery through final guidelines.

Commands:

Discovery Phase

Strategy Phase

Messaging Phase

Visual Phase

Compile Phase

Agents:

• brand-archetype-selector - Jungian archetype selection
• brand-audience-researcher - Psychographic analysis and audience profiling
• brand-color-strategist - Strategic color palette development
• brand-competitive-auditor - Competitor visual identity analysis
• brand-elevator-pitch-writer - Elevator pitch variations
• brand-messaging-architect - Value proposition and brand pillars
• brand-naming-specialist - Strategic name development
• brand-positioning-strategist - Positioning and onliness statements
• brand-purpose-architect - Purpose, mission, vision (Golden Circle)
• brand-tagline-creator - Memorable tagline creation
• brand-typography-curator - Typography system and font selection
• brand-values-curator - Core values discovery and articulation
• brand-visual-director - Visual identity direction and briefs
• brand-voice-architect - Voice and verbal identity guidelines
• startup-brand-architect - Orchestrates complete brand identity creation

🧐 01-START (Discovery & Planning)

Plan before you build. Discover the problem space, scope your MVP, and create an implementation roadmap.

Commands:

Agents:

• market-validator - Research market viability and competition
• feature-brainstormer - Generate and prioritize feature ideas
• tech-advisor - Recommend technology choices
• plan-reviewer - Review and improve implementation plans
• data-modeler - Design data structures and schemas

🧰 TOOLKIT (Specialized Tools)

Utility commands and agents for tasks outside the main development workflow.

Commands:

Agents:

• brand-researcher - Deep brand research for content creators
• business-growth-advisor - Strategic business growth consultation
• deep-researcher - Comprehensive expert knowledge research
• elite-copywriter - Transform text into compelling copy
• hook-generator - Generate scroll-stopping hook variations
• nano-banana-pro-expert - Guidance for Nano Banana Pro image generation
• sponsor-script-writer - Write high-converting sponsored content scripts

🏗️ 02-BUILD (Implementation)

Build features methodically. Plan each feature, implement with best practices, and review before shipping.

Commands:

Agents:

• code-architect - Design feature architecture and patterns
• code-guru - Implement features with best practices
• code-reviewer - Thorough code review and quality checks
• tester - Write and run tests iteratively to prove code works

🚀 03-SHIP (Deployment)

Ship with confidence. Run checks, commit cleanly, and create PRs.

Commands:

🔧 04-DEBUG (Debugging)

Fix bugs systematically. Diagnose root causes, apply fixes, and verify they work.

Commands:

Agents:

• diagnostician - Deep investigation and root cause analysis
• fixer - Minimal, targeted bug fixes
• verifier - Verification and regression testing
• tester - Write and run tests to prove fixes work

✨ 05-REFACTOR (Code Evolution)

Improve code without changing behavior. Assess opportunities, refactor safely, and validate preservation.

Commands:

Agents:

• assessor - Code archaeology and improvement analysis
• refactorer - Safe structural improvements
• validator - Behavior preservation verification
• tester - Write and run tests to prove behavior preserved

🧠 Skills

Skills are knowledge packs that Claude loads on its own when a conversation calls for them. You never invoke them by name. The commands and agents above lean on them, and they also work in any ordinary conversation once the plugin is installed.

platform-optimization and midjourney-prompting describe fast-moving targets. Each carries a last-verified date and tells Claude to check current sources before relying on specific numbers.

Building your own skills, agents, hooks, or plugins? Version 1 shipped builder skills for that. Anthropic now maintains better ones alongside the product, so install plugin-dev and skill-creator from the official marketplace (/plugin marketplace add anthropics/claude-plugins-official).

🔌 MCP Servers

MCP (Model Context Protocol) servers extend Claude Code with additional capabilities. This plugin suite auto-installs essential servers that work out of the box.

✅ Auto-Installed (No Setup Required)

These servers start automatically when the plugin is enabled:

Usage Tips:

• Add use context7 to prompts for current library documentation
• Cross-session learning uses Claude Code's native agent memory (stored in .claude/agent-memory/ in your project), so no memory server is needed
• Taskmaster config is created automatically during installation (step 4)

⚙️ Optional Servers (Require Setup)

Add these based on your workflow. Each requires authentication or additional setup.

🔀 Version Control & Issues

🐛 Debugging & Testing

🗄️ Database & Backend

🌐 Deployment

💡 Best Practice

Start with the auto-installed essentials. Add specialized servers only when you need them. Too many MCP servers can slow down Claude Code startup.

📁 Structure

claude-vibes/
├── .claude-plugin/
│   └── marketplace.json         # Plugin marketplace manifest
├── README.md                    # This file
└── plugins/
    └── vibes/                   # Bundled plugin
        ├── .claude-plugin/
        │   └── plugin.json      # Plugin manifest with MCP servers
        ├── commands/
        │   ├── 00-BRAND/        # Brand Identity
        │   ├── 01-START/        # Discovery & Planning
        │   ├── 02-BUILD/        # Implementation
        │   ├── 03-SHIP/         # Deployment
        │   ├── 04-DEBUG/        # Debugging
        │   ├── 05-REFACTOR/     # Code Evolution
        │   └── TOOLKIT/         # Specialized Tools
        ├── agents/
        │   ├── BRANDING/        # Brand identity agents
        │   ├── CODING/          # Development agents
        │   └── TOOLKIT/         # Utility agents
        └── skills/              # Agent enhancement skills

📚 References

Plugins

• Plugins Overview - Install and create plugins
• Plugins Reference - Technical specifications
• Plugin Announcement - Introduction to the plugin system

Plugin Components

• Slash Commands - Custom command syntax
• Subagents - Specialized agent configuration
• Hooks Reference - Event-driven automation
• Agent Skills - Model-invoked capabilities

MCP (Model Context Protocol)

• Claude Code MCP - Connect to external tools
• Introducing MCP - Protocol overview

Claude Code

• Claude Code Overview - Getting started
• Best Practices - Agentic coding patterns