AgentStudio is a modern, web-based personal agent workspace platform built on top of Claude Code SDK. It extends Claude Code's capabilities with a professional web interface, making AI-powered development accessible to everyone.
- Professional, intuitive web UI designed for both developers and general users
- Real-time streaming responses for immediate feedback
- Split-panel layout with chat interface and live preview
- Responsive design optimized for desktop and mobile
- Claude Integration: Built on Claude Code SDK with full API support
- Multiple Providers: Support for OpenAI, GLM, DeepSeek, and other popular LLMs
- Flexible Configuration: Easy switching between different AI models
- Streaming Input: Real-time response streaming for better user experience
- Built-in Agents: PPT Editor, Code Assistant, Document Writer
- Custom Agents: Create and configure your own specialized agents
- Subagents: User-defined AI agents with custom system prompts
- Agent Marketplace: Extensible agent development framework
- Message-level Tool Customization: Fine-grained control over agent capabilities
- Built-in File Browser: Navigate and manage project files seamlessly
- File Content Viewer: Preview and edit files directly in the interface
- Project-aware Operations: Context-sensitive file operations
- Version Control Integration: Git-aware file management
- Slide Agent: Create and edit presentations with AI assistance
- Code Explorer: Navigate codebases with intelligent search
- Document Outline: Structured document editing and management
- Image Upload: Support for visual content analysis
- Tool Renderer: Dynamic visualization of tool usage and results
- TypeScript Throughout: Full type safety for better development experience
- Modern Stack: React 18, Vite, TailwindCSS, Zustand
- Testing Suite: Comprehensive test coverage with Vitest
- Hot Reload: Fast development with instant code updates
- Local Development: Easy setup for development environments
- One-Click Installation: Simplified deployment for end users
- Production Builds: Optimized builds for performance
- Cross-Platform: Support for Linux, macOS, and Windows
One-Command Setup:
# Build and run with docker-compose
docker build -t agentstudio:latest .
docker-compose up -dThen open http://localhost in your browser.
What you get:
- β All-in-One Container: Frontend + Backend in a single container
- β Data Persistence: Automatic data backup via Docker volumes
- β Zero Configuration: Works out of the box
- β Easy Updates: Simple rebuild and restart
π Full Documentation:
- Docker Deployment Guide - Detailed setup and configuration
- Quick Start Guide - Step-by-step tutorial with examples
π§ Linux & π macOS - User Installation (Recommended - No sudo required)
# Install Agent Studio backend in user space
curl -fsSL https://raw.githubusercontent.com/okguitar/agentstudio/main/scripts/remote-install.sh | bashThe installer will:
- β Auto-detect and install Node.js 18+ (via system package manager or NVM)
- β Auto-install pnpm for faster package management (optional)
- β Handle all dependencies automatically
- β Support all major Linux distributions (Ubuntu, CentOS, Fedora, Arch, etc.)
- β Work with both root and regular users
The installer will ask if you want to start the backend immediately. If you choose not to start now, you can start it later:
# Start the backend
~/.agent-studio/start.sh
# Stop the backend
~/.agent-studio/stop.shπͺ Windows - PowerShell Installation
# Run as Administrator in PowerShell
PowerShell -ExecutionPolicy Bypass -Command "iex ((New-Object System.Net.WebClient).DownloadString('https://raw.githubusercontent.com/okguitar/agentstudio/main/scripts/windows-install.ps1'))"The Windows installer will:
- β Auto-install Node.js (via Chocolatey, winget, or direct download)
- β Auto-install Git if not available
- β Handle all dependencies automatically
- β Create start/stop batch scripts
Windows Alternative - Simple Batch Script If you already have Node.js and Git installed:
# Download and run the simple installer
curl -o windows-install-simple.bat https://raw.githubusercontent.com/okguitar/agentstudio/main/scripts/windows-install-simple.bat
windows-install-simple.batAccess the application:
- Open your browser and visit: https://agentstudio-frontend.vercel.app/
- In the web interface, go to Settings β API Configuration
- Enter your backend URL:
http://localhost:4936 - Click "Test Connection" to verify
Option 2: System Service Installation (Requires sudo)
For production deployments with automatic startup on boot:
# Install as system service with auto-start
curl -fsSL https://raw.githubusercontent.com/okguitar/agentstudio/main/scripts/remote-install.sh | sudo bashAfter installation, manage the service:
agent-studio start # Start the service
agent-studio stop # Stop the service
agent-studio restart # Restart the service
agent-studio status # Check service status
agent-studio logs # View logs
agent-studio config # Edit configurationThen access the web interface at https://agentstudio-frontend.vercel.app/ and configure the backend URL in Settings.
Prerequisites:
- Node.js 18+
- pnpm (recommended) or npm
- Git
Installation:
# Clone the repository
git clone https://github.com/okguitar/agentstudio.git
cd agentstudio
# Install dependencies
pnpm install
# Start development servers (both frontend and backend)
pnpm run dev
# Or start them separately
pnpm run dev:frontend # Frontend only (port 3000)
pnpm run dev:backend # Backend only (port 4936)Build for production:
pnpm run build
pnpm startagentstudio/
βββ frontend/ # React frontend application
β βββ src/
β β βββ components/ # Reusable UI components
β β βββ pages/ # Page components
β β βββ agents/ # Agent-specific components
β β βββ hooks/ # React hooks and data fetching
β β βββ stores/ # State management (Zustand)
β βββ public/ # Static assets
βββ backend/ # Node.js backend server
β βββ src/
β β βββ routes/ # API endpoints
β β βββ services/ # Business logic
β β βββ index.ts # Server entry point
β βββ dist/ # Built backend code
βββ shared/ # Shared types and utilities
βββ scripts/ # Installation and deployment scripts
For Development:
Create a .env file in the backend/ directory:
# AI Provider Configuration (choose one or more)
ANTHROPIC_API_KEY=your_anthropic_api_key_here
OPENAI_API_KEY=your_openai_api_key_here
# Server Configuration
PORT=4936
NODE_ENV=development
# File System
SLIDES_DIR=../slides
# CORS Configuration (for production)
CORS_ORIGINS=https://your-frontend-domain.comFor System Service Installation:
Edit the configuration file at /etc/agent-studio/config.env:
# Edit configuration
agent-studio configConfiguration options:
# Server configuration
NODE_ENV=production
PORT=4936
SLIDES_DIR=/opt/slides
# Optional: AI provider keys (configure if needed)
# ANTHROPIC_API_KEY=your_anthropic_api_key_here
# OPENAI_API_KEY=your_openai_api_key_here
# Optional: CORS configuration for custom frontends
# CORS_ORIGINS=https://your-frontend.vercel.app,https://custom-domain.comThe frontend is hosted at https://agentstudio-frontend.vercel.app/ and can connect to different backend instances:
For Local Backend:
- Open Settings β API Configuration in the web interface
- Enter:
http://localhost:4936(or your custom port) - Click "Test Connection" to verify
For Remote Backend:
- Deploy your backend to a server with a public IP or domain
- Configure the backend URL in Settings (e.g.,
https://your-backend.com) - Ensure CORS is properly configured in your backend's
.envfile
# Basic service operations
agent-studio start # Start the service
agent-studio stop # Stop the service
agent-studio restart # Restart the service
agent-studio status # Check service status
# Monitoring and logs
agent-studio logs # View real-time logs
agent-studio health # Check service health
# Configuration
agent-studio config # Edit configuration file
# Service details
agent-studio uninstall # Remove the serviceInstallation Directories:
For user installation (no sudo):
- Application:
~/.agent-studio - Configuration:
~/.agent-studio-config/config.env - Logs:
~/.agent-studio-logs/ - Slides:
~/slides - Start script:
~/.agent-studio/start.sh - Stop script:
~/.agent-studio/stop.sh
For system service installation (with sudo):
- Application:
~/.agent-studio - Configuration:
~/.agent-studio-config/config.env - Logs:
~/.agent-studio-logs/ - Slides:
~/slides
Log Files:
- Output log:
/var/log/agent-studio/output.log - Error log:
/var/log/agent-studio/error.log - Log rotation: Daily (30 days retention)
Service Integration:
- Linux: systemd service (
/etc/systemd/system/agent-studio.service) - macOS: launchd service (
/Library/LaunchDaemons/com.agent-studio.backend.plist) - Auto-start on boot enabled by default
Node.js Installation Problems: The installer automatically handles Node.js installation, but if you encounter issues:
# For Linux/macOS - Manual Node.js installation via NVM
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.0/install.sh | bash
source ~/.bashrc
nvm install --lts
nvm use --lts
# For Windows - Download from official website
# Visit: https://nodejs.org/Ubuntu/Debian Specific Issues:
- Root User Detection: Fixed in latest installer - now properly supports both root and regular users
- TTY Issues: Installer sets
CI=trueto handle non-interactive environments - Build Failures: Installer automatically retries with dev dependencies
Windows Installation Issues:
- PowerShell Execution Policy: Run
Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser - Missing Dependencies: Installer will prompt to install Node.js, Git, and other requirements
- Permission Issues: Run PowerShell as Administrator for system-wide installations
For User Installation:
# Check if backend is running
curl http://localhost:4936/api/health
# View logs
cat ~/.agent-studio-logs/output.log
cat ~/.agent-studio-logs/error.log
# Restart backend
~/.agent-studio/stop.sh
~/.agent-studio/start.shFor System Service Installation:
# Check service status
agent-studio status
# View logs for errors
agent-studio logs
# Verify configuration
agent-studio config
# Check if port is available
lsof -i :4936Permission Errors (User Installation):
# Fix script permissions
chmod +x ~/.agent-studio/start.sh
chmod +x ~/.agent-studio/stop.sh
# Fix directory permissions
chmod -R 755 ~/.agent-studioPermission Errors (System Service):
sudo chown -R agent-studio:agent-studio /opt/agent-studio
sudo chown -R agent-studio:agent-studio /var/log/agent-studioPort Already in Use:
# Find process using the port
lsof -i :4936
# Kill the process (replace PID with actual process ID)
kill -9 <PID>
# Or change port in configuration
# Edit ~/.agent-studio-config/config.env and change PORT=8080Build Failures: The installer now automatically handles build failures by:
- Installing dev dependencies
- Retrying the build
- Falling back to development mode if build still fails
Health Check:
# For user installation
curl http://localhost:4936/api/health
# For system service
agent-studio healthWindows Specific Issues:
REM Check if Node.js is available
node --version
REM Check if backend is running
curl http://localhost:4936/api/health
REM Start backend manually
cd %USERPROFILE%\.agent-studio
start.bat
REM Stop backend
stop.batTo update an existing installation:
# Stop the service
agent-studio stop
# Update the code
cd /opt/agent-studio
git pull
pnpm install
pnpm run build:backend
# Start the service
agent-studio start# Run all tests
pnpm test
# Run frontend tests only
cd frontend && pnpm test
# Run tests with coverage
cd frontend && pnpm run test:coverage
# Run tests with UI
cd frontend && pnpm run test:ui- API Key Protection: Environment-based API key management
- CORS Configuration: Configurable cross-origin policies
- Input Validation: Comprehensive input sanitization
- Secure File Operations: Sandboxed file system access
We welcome contributions! Please see our Contributing Guidelines for details.
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
This project is licensed under the MIT License - see the LICENSE file for details.
- Built on top of Claude Code SDK
- Inspired by modern development workflows
- Community feedback and contributions
| link | type | service | description |
|---|---|---|---|
| ctok.ai | π€ community | β
Claude Code β Codex CLI |
Claude Code / Codex CLI carpool service. |
θ―·ζ₯η README.zh-CN.md θ·εδΈζζζ‘£γ
