Claudex

Your own Claude Code UI. Open source, self-hosted, runs entirely on your machine.

Community

Join our Discord server to get help, share feedback, and connect with other users.

Why Claudex?

• Multiple sandboxes - Docker (local), E2B (cloud), or Modal (cloud).
• Use your own plans - Claude Max, OpenRouter, OpenAI, or custom providers.
• Full IDE experience - VS Code in browser, terminal, file explorer.
• Extensible - Skills, agents, slash commands, MCP servers.

Screenshots

Quick Start

git clone https://github.com/Mng-dev-ai/claudex.git
cd claudex
docker compose up -d

Open http://localhost:3000

Desktop (macOS)

See Desktop Setup to run Claudex as a native macOS app.

Features

Sandboxed Code Execution

Run AI agents in isolated environments with multiple sandbox providers:

• Docker - Fully local, no external dependencies
• E2B - Cloud sandboxes with e2b.dev
• Modal - Serverless cloud sandboxes with modal.com

Full Development Environment

• VS Code editor in the browser
• Terminal with full PTY support
• File system management
• Port forwarding for web previews
• Environment checkpoints and snapshots

VNC Browser Control

View and interact with a browser running inside the sandbox via VNC. Use Playwright MCP with Chrome DevTools Protocol (CDP) to let Claude control the browser programmatically.

Multiple AI Providers

Switch between providers in the same chat:

• Anthropic - Use your Max plan
• OpenAI - Use your ChatGPT Pro subscription (GPT-5.2 Codex, GPT-5.2)
• OpenRouter - Access to multiple model providers
• Custom - Any Anthropic-compatible API endpoint

Extend with Skills & Agents

• Custom Skills - ZIP packages with YAML metadata
• Custom Agents - Define agents with specific tool configurations
• Slash Commands - Built-in (/context, /compact, /review, /init)
• MCP Servers - Model Context Protocol support (NPX, BunX, UVX, HTTP)

Scheduled Tasks

Automate recurring tasks with Celery workers.

Chat Features

• Fork chats from any message point
• Restore to any previous message in history
• File attachments with preview

Preview Capabilities

• Web preview for running applications
• Mobile viewport simulation
• File previews: Markdown, HTML, images, CSV, PDF, PowerPoint

Marketplace

• Browse and install plugins from official catalog
• One-click installation of agents, skills, commands, MCPs

Secrets Management

• Environment variables for sandbox execution

Integrations

• Gmail - Read, send, and manage emails via Gmail MCP Server

Gmail Setup

1. Go to Google Cloud Console
2. Create a project and enable the Gmail API
3. Create OAuth credentials (Desktop app for localhost, Web application for hosted URLs)
4. If using Web application, add redirect URI: https://YOUR_DOMAIN/api/v1/integrations/gmail/callback
5. Download the JSON credentials file
6. In Claudex Settings → Integrations, upload your credentials file
7. Click Connect Gmail to authorize

Custom Instructions

• System prompts for global context
• Custom instructions injected with each message

Configuration

Configure providers in Settings → Providers after login.

All providers use Claude Code under the hood. Non-Anthropic providers work through Anthropic Bridge, which translates Anthropic API calls to other providers.

┌─────────────┐     ┌───────────────────┐     ┌───────────────────────┐
│   Claudex   │────▶│  Anthropic Bridge │────▶│  OpenAI / OpenRouter  │
│             │     │  (API Translator) │     │  / Custom             │
└─────────────┘     └───────────────────┘     └───────────────────────┘

This means all providers share the same conversation history stored in ~/.claude JSONL files, plus the same slash commands, skills, agents, and MCP servers. You can develop a feature with Claude, then switch to GPT-5.2 Codex for review—it already has the full context without needing to re-read files.

Supported Providers

Provider	Auth Method	Models
Anthropic	OAuth token from claude setup-token	Sonnet 4.5, Opus 4.5, Haiku 4.5
OpenAI	Auth file from codex login	GPT-5.2 Codex, GPT-5.2
OpenRouter	API key	Multiple providers
Custom	API key	Any Anthropic-compatible endpoint

OpenAI Setup (ChatGPT Pro)

Use OpenAI models with your ChatGPT Pro subscription:

1. Install Codex CLI
2. Run codex login and authenticate with your ChatGPT account
3. In Claudex Settings → Providers, add an OpenAI provider
4. Upload your ~/.codex/auth.json file

Custom Providers

Use any Anthropic-compatible API endpoint:

1. Get your API key from your provider
2. In Claudex Settings → Providers, add a Custom provider
3. Enter your API endpoint URL and API key

Compatible coding plans:

• GLM Coding Plan
• Kimi Coding Plan
• MiniMax Coding Plan

You only need one AI provider configured.

Architecture

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│    Frontend     │────▶│   FastAPI       │────▶│   PostgreSQL    │
│   React/Vite    │     │   Backend       │     │   Database      │
└─────────────────┘     └────────┬────────┘     └─────────────────┘
                                 │
                    ┌────────────┼────────────┐
                    ▼            ▼            ▼
            ┌───────────┐ ┌───────────┐ ┌─────────────────┐
            │   Redis   │ │  Celery   │ │ Docker Sandbox  │
            │  Pub/Sub  │ │  Workers  │ │                 │
            └───────────┘ └───────────┘ └─────────────────┘

Tech Stack

Frontend: React 19, TypeScript, Vite, TailwindCSS, Zustand, React Query, Monaco Editor, XTerm.js

Backend: FastAPI, Python 3.13, SQLAlchemy 2.0, Celery, Redis, Granian

Services

Service	Port
Frontend	3000
Backend API	8080
PostgreSQL	5432
Redis	6379

Commands

docker compose up -d      # Start
docker compose down       # Stop
docker compose logs -f    # Logs

Deployment

For production deployment on a VPS, see the Coolify Installation Guide.

API & Admin

• API Docs: http://localhost:8080/api/v1/docs
• Admin Panel: http://localhost:8080/admin

Default admin: admin@example.com / admin123

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Open a Pull Request

License

Apache 2.0 - see LICENSE