---
title: Claude Kit
description: The open-source AI dev toolkit for Claude Code. 44 skills, 20 agents, 7 modes — install as a plugin and go. Free forever.
template: splash
hero:
  tagline: The open-source AI dev toolkit for Claude Code. 44 skills, 20 agents, 7 modes — install as a plugin and go. Free forever.
  image:
    dark: ../../assets/hero-dark.svg
    light: ../../assets/hero-light.svg
  actions:
    - text: Get Started
      link: /claudekit/getting-started/introduction/
      icon: right-arrow
      variant: primary
    - text: View on GitHub
      link: https://github.com/duthaho/claudekit
      icon: external
      variant: minimal
---

import { Card, CardGrid, LinkCard } from '@astrojs/starlight/components';

## Ship Faster. For Free.

Claude Kit transforms Claude Code into a production-ready AI development team. Pre-built skills, specialized agents, and intelligent modes — all open source.

<CardGrid>
  <Card title="44 Skills" icon="puzzle">
    Auto-triggered knowledge for Python, TypeScript, React, FastAPI, TDD, debugging, security, and more. Claude knows your stack without being told.
  </Card>
  <Card title="20 Agents" icon="rocket">
    Specialized subagents for code review, testing, database design, security auditing, UI/UX, and more — each with focused expertise.
  </Card>
  <Card title="7 Modes" icon="setting">
    Switch between brainstorm, implementation, review, deep research, and more. Each mode optimizes Claude's behavior for your task.
  </Card>
  <Card title="MCP Servers" icon="external">
    Real-time docs via Context7, persistent memory, browser testing, and structured reasoning via Model Context Protocol.
  </Card>
</CardGrid>

---

## Quick Start

Get up and running in under 2 minutes:

```bash
# Add the marketplace
/plugin marketplace add duthaho/claudekit-marketplace

# Install the plugin
/plugin install claudekit

# Configure your project (optional)
/claudekit:init
```

Skills trigger automatically based on what you're doing. Ask Claude to brainstorm a feature, write a plan, debug an error, or review code — the right skills activate without any commands needed.

---

## Why Claude Kit?

### The Problem

Working with Claude Code is powerful, but raw Claude has limitations:

- **Context Spirals** — Token budgets run out, context gets lost mid-task
- **Inconsistent Output** — Quality varies between sessions
- **No Structure** — Every session starts from scratch
- **Missing Expertise** — Claude doesn't know your team's patterns and standards

### The Solution

Claude Kit provides the structure, patterns, and automation that makes Claude Code production-ready:

<CardGrid>
  <Card title="Auto-Triggered Skills" icon="puzzle">
    Say "fix this bug" and the systematic-debugging skill activates. Say "plan this feature" and brainstorming + writing-plans kick in. No commands to memorize.
  </Card>
  <Card title="Specialized Agents" icon="rocket">
    Delegate code review, security audits, testing, and documentation to focused subagents that work in parallel.
  </Card>
  <Card title="Consistent Quality" icon="approve-check">
    Built-in TDD enforcement, verification before completion, and code review workflows ensure every output meets your standards.
  </Card>
  <Card title="Fully Customizable" icon="pencil">
    Add your own skills, agents, and modes. Claude Kit is a template, not a black box.
  </Card>
</CardGrid>

---

## How It Works

Claude Kit uses three layers that work together:

<CardGrid>
  <LinkCard
    title="Skills"
    description="44 knowledge modules that auto-trigger based on keywords. Cover languages, frameworks, testing, debugging, security, and development methodology."
    href="/claudekit/reference/skills/"
  />
  <LinkCard
    title="Agents"
    description="20 specialized subagents you can dispatch for focused tasks: code review, testing, database design, security audits, and more."
    href="/claudekit/reference/agents/"
  />
  <LinkCard
    title="Modes"
    description="7 behavioral configurations that change how Claude communicates and solves problems: brainstorm, implementation, review, and more."
    href="/claudekit/reference/modes/"
  />
  <LinkCard
    title="MCP Servers"
    description="Optional integrations for real-time library docs, persistent memory, browser automation, and structured reasoning."
    href="/claudekit/reference/mcp-servers/"
  />
</CardGrid>

---

## Development Workflows

Claude Kit connects skills and agents into complete workflows:

| Workflow | What Happens |
|----------|-------------|
| **Planning & Building** | Brainstorm requirements, write implementation plans, execute with subagents |
| **Testing & Debugging** | TDD enforcement, systematic debugging, root cause tracing, verification |
| **Reviewing & Shipping** | Code review, git workflows, PR creation, changelog generation |

[Explore workflows →](/claudekit/workflows/planning-and-building/)

---

## Ready to Ship Faster?

<CardGrid>
  <LinkCard
    title="Installation Guide"
    description="Step-by-step setup for your project"
    href="/claudekit/getting-started/installation/"
  />
  <LinkCard
    title="Skills Reference"
    description="All 44 skills organized by category"
    href="/claudekit/reference/skills/"
  />
  <LinkCard
    title="Agents Reference"
    description="All 20 specialized subagents"
    href="/claudekit/reference/agents/"
  />
  <LinkCard
    title="Customization"
    description="Create your own skills, agents, and modes"
    href="/claudekit/customization/creating-skills/"
  />
</CardGrid>