One Command Setup

Set Up Claude Code In 60 Seconds

Name: Claude Code Setup Tool
Author: TunerLabs

One curl command installs a smart Claude Skill that auto-detects your stack, asks targeted questions, and generates a perfectly tailored Claude Code environment for any project.

curl -sL https://tunerlabs.com/tools/claude-code-setup/install.sh | bash

Works on macOS & Linux · Requires Claude Code CLI · Restart Claude Code after install

Watch the full Claude Code walkthrough

📚 Claude Code Series

The Complete Architecture Reference

Master Skills, Agents, Memory hierarchy, MCP servers, and every tool in one interactive guide.

Read Guide → 02

The Four-Layer Memory Stack

Make Claude Code remember everything across sessions with 4 simple files.

Read Guide →

One-Command Setup Tool

You are here — auto-configure Claude Code for any project in 60 seconds.

✅ Currently reading

Getting Started

Three Steps. That's It.

Install once globally, then use in every project forever. Claude handles the intelligence — no templates, no boilerplate.

Install the Skill

Run one command. It installs a global Claude Skill to ~/.claude/skills/ — available in every project. Restart Claude Code after installing.

curl -sL tunerlabs.com/tools/claude-code-setup/install.sh | bash

Run in Your Project

Start a new Claude Code session in any project and type the slash command. Claude starts analyzing your codebase immediately.

> /setup-claude-code

Answer & Done

Claude asks a few targeted questions about your preferences, then generates every config file tailored to your actual project.

✓ 8 files generated · ready to code

    ⚡
    Why not just a template? Templates give generic boilerplate. This tool uses Claude's intelligence to read your actual code — it discovers your real directory structure, real dependencies, real scripts, and generates instructions that reference your actual project. The difference: "use pnpm" vs "use pnpm, components are in src/app/Components/ with capital C convention, Tailwind v4 via PostCSS."
  

Output

Everything It Generates

The tool can generate up to 8 categories of configuration files, each tailored to your project's actual stack and structure.

📄

CLAUDE.md

Project memory — stack info, commands, architecture patterns, conventions, and anti-patterns. Claude reads this before every session.

Project

🔒

.claude/settings.json

Permission rules — what Claude can read, write, and execute. Scoped by paths and commands with allow/deny lists.

Project

⚙️

.claude/rules/*.md

Modular convention rules — path-scoped for monorepos. API rules for backend, component rules for frontend.

Project

🛠️

.claude/skills/*/SKILL.md

Project-specific skills — deploy workflows, seed database commands, migration generators. Invokable via slash commands.

Project

🤖

.claude/agents/*.md

Specialist agents — security reviewer, DB migration expert, API designer. Each with scoped tools and tailored prompts.

Project

🔗

.claude/mcp.json

MCP server configs — GitHub, PostgreSQL, Slack, and more. Pre-configured with the right transport and commands.

Project

👤

~/.claude/CLAUDE.md

Global personal preferences — your default package manager, code style, commit format. Applies to all projects.

Global

🔧

~/.claude/rules/*.md

Global rules — git conventions, testing patterns, formatting standards. Modular files you set once and forget.

Global

Intelligence

What It Auto-Detects

The tool reads your project before asking anything. Most configuration is derived automatically — you only answer what can't be detected.

🔍 Language & Framework

package.json → Node.js, React, Next.js, Vue, Svelte
pyproject.toml → Python, Django, Flask, FastAPI
go.mod → Go modules and dependencies
Cargo.toml → Rust crates and edition
Gemfile → Ruby, Rails version

📦 Package Manager & Build

pnpm-lock.yaml → pnpm detected
yarn.lock → Yarn detected
package-lock.json → npm detected
Scripts from package.json → test, build, dev commands
Build configs → Vite, Webpack, esbuild, tsconfig

🗃️ Database & ORM

Prisma, Drizzle, TypeORM, Sequelize detection
SQLAlchemy, Django ORM, ActiveRecord
Diesel (Rust), GORM (Go), Ecto (Elixir)
Database type from connection configs

🚀 Project Architecture

pnpm-workspace.yaml → monorepo detected
nx.json / turbo.json → build orchestration
Directory structure analysis (src/, app/, lib/, packages/)
CI/CD from .github/workflows/ or .gitlab-ci.yml
Lint & format from .eslintrc, biome.json, .prettierrc

Interaction

What It Asks You

Only questions that require human judgment. Everything else is auto-detected. The tool pre-fills answers from detection and lets you confirm or customize.

Setup Scope

Configure just this project, or also set up your global personal preferences?

Project only Project + Global Global only

Code Patterns

What patterns should Claude always follow? Pre-filled with detected conventions from your codebase.

Named exports only Repository pattern Custom...

Anti-Patterns

What should Claude never do? Suggested based on your detected stack.

No default exports No any type No console.log

Permission Strictness

How locked-down should Claude's tool access be?

Relaxed Moderate Strict

MCP Servers & Extras

Configure MCP servers, starter skills, and project-specific agents based on what was detected.

GitHub MCP PostgreSQL MCP Deploy skill Security agent

Transformation

Before & After

Your project goes from zero Claude configuration to a fully tailored environment in one command.

Before

Project Root
my-project/├──package.json├──src/│   ├──app/│   └──lib/└──tsconfig.json  No CLAUDE.md  No permissions  No skills or agents  Claude knows nothing about your project

After /setup-claude-code

Project Root
my-project/├──CLAUDE.md← project memory├──CLAUDE.local.md← personal overrides├──.claude/│   ├──settings.json← permissions│   ├──rules/│   │   ├──api-patterns.md│   │   └──frontend-rules.md│   ├──skills/│   │   └──Deploy/│   │       └──SKILL.md│   ├──agents/│   │   └──security-reviewer.md│   └──mcp.json← MCP servers├──package.json├──src/└──tsconfig.json

Compatibility

Works With Every Stack

Auto-detection supports all major languages, frameworks, and toolchains. The generated config is always specific to what you're actually using.

JavaScript

TypeScript

Python

Rust

Java

Ruby

PHP

.NET / C#

Kotlin

Swift

Dart / Flutter

Frameworks & Tools

React

Next.js

Vue

Svelte

Django

Flask

FastAPI

Rails

Express

GraphQL

Docker

Kubernetes

PostgreSQL

MongoDB

Prisma

Drizzle

Jest

Vitest

pytest

pnpm

Yarn

npm

Deep Dive

Want to Understand the Full Architecture?

The setup tool generates configs based on our complete Claude Code architecture guide. Read it to understand the 7-level memory hierarchy, how skills work, agent design patterns, and the full tools reference.

Read the Full Guide → Install the Tool ↑

    💡
    Pro tip: Run /setup-claude-code again anytime — if your project is already configured, it shows a status dashboard and lets you add new skills, agents, rules, or MCP servers without redoing the full setup.
  

Ready to Set Up Claude Code?

curl -sL https://tunerlabs.com/tools/claude-code-setup/install.sh | bash

Restart Claude Code after install · Then run /setup-claude-code in any project