Orgo MCP Server
An MCP (Model Context Protocol) server that gives AI agents the ability to control virtual computers through Orgo.
Quick Start (Cloud-Hosted)
The fastest way to use Orgo MCP - no installation required!
1. Get Your API Key
Sign up at orgo.ai and copy your API key.
2. Configure Your Client
claude mcp add --transport http orgo https://orgo-mcp.onrender.com/mcp \
--header "X-Orgo-API-Key: YOUR_API_KEY"
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"orgo": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://orgo-mcp.onrender.com/mcp",
"--header", "X-Orgo-API-Key:YOUR_API_KEY"]
}
}
}
Note: No space after the colon in X-Orgo-API-Key:YOUR_API_KEY (mcp-remote quirk).
For team sharing, create .mcp.json at your project root:
{
"mcpServers": {
"orgo": {
"type": "http",
"url": "https://orgo-mcp.onrender.com/mcp",
"headers": {
"X-Orgo-API-Key": "${ORGO_API_KEY}"
}
}
}
}
Each team member sets their own key: export ORGO_API_KEY="sk_live_xxx"
3. Start Using
"Create a Linux computer with 4GB RAM"
"Take a screenshot of my computer"
"Run 'ls -la' on the computer"
"Type 'hello world' and press Enter"
Features (34 tools)
Category | Tools |
Projects | orgo_list_projects
, orgo_create_project
, orgo_get_project
, orgo_delete_project
, orgo_start_project
, orgo_stop_project
, orgo_restart_project
|
Computers | orgo_list_computers
, orgo_create_computer
, orgo_get_computer
, orgo_start_computer
, orgo_stop_computer
, orgo_restart_computer
, orgo_delete_computer
|
Actions | orgo_screenshot
, orgo_click
, orgo_double_click
, orgo_type
, orgo_key
, orgo_scroll
, orgo_drag
, orgo_wait
|
Shell | orgo_bash
, orgo_exec
|
Files | orgo_list_files
, orgo_upload_file
, orgo_export_file
, orgo_download_file
, orgo_delete_file
|
Streaming | orgo_start_stream
, orgo_stream_status
, orgo_stop_stream
|
AI | orgo_list_ai_models
, orgo_ai_completion
|
Self-Hosting Options
Option 1: Local Development (stdio)
git clone https://github.com/nickvasilescu/orgo-mcp.git
cd orgo-mcp
pip install -e .
export ORGO_API_KEY="your_key"
python orgo_mcp.py
Add to claude_desktop_config.json:
{
"mcpServers": {
"orgo": {
"command": "python3",
"args": ["/path/to/orgo-mcp/orgo_mcp.py"],
"env": {"ORGO_API_KEY": "your_key"}
}
}
}
Option 2: Local HTTP Server
pip install -e .
MCP_TRANSPORT=http python orgo_mcp.py
Server available at http://localhost:8000/mcp
Option 3: Docker
# Build
docker build -t orgo-mcp .
# Run
docker run -p 8000:8000 orgo-mcp
# Or use docker-compose
docker-compose up
Option 4: Deploy to Render.com
Fork this repository
Go to Render Dashboard
Click "New Blueprint Instance"
Connect your GitHub repo
Deploy!
Your server will be at: https://orgo-mcp-xxxx.onrender.com
Option 5: Deploy to Fly.io
# Install flyctl
curl -L https://fly.io/install.sh | sh
# Login and deploy
fly auth login
fly launch --no-deploy
fly deploy
Your server will be at: https://orgo-mcp.fly.dev
Environment Variables
Variable | Default | Description |
MCP_TRANSPORT
| stdio
| Transport mode: stdio
or http
|
MCP_HOST
| 0.0.0.0
| HTTP bind address |
MCP_PORT
/ PORT
| 8000
| HTTP port |
ORGO_API_KEY
| - | API key (required for stdio transport) |
CORS_ORIGINS
| *
| Allowed origins (comma-separated) |
Usage Examples
Project Management
"Create a new project called 'qa-automation'"
"List my Orgo projects"
"Start all computers in project proj_123"
"Delete project proj_123"
Computer Management
"Create a new Linux computer called 'dev-box' with 4GB RAM"
"Get details for computer abc123"
"Start computer abc123"
"Stop computer abc123"
Screen Actions
"Take a screenshot of computer abc123"
"Click at coordinates (500, 300)"
"Type 'hello world'"
"Press Enter"
"Scroll down"
Shell Commands
"Run 'ls -la' on computer abc123"
"Execute Python code: print('hello')"
File Operations
"List files on computer abc123"
"Export file ~/Documents/report.pdf from computer abc123"
"Upload file to computer abc123"
Streaming
"Start streaming computer abc123 to Twitch"
"Check stream status for computer abc123"
"Stop streaming computer abc123"
AI Completion
"List available AI models"
"Run GPT-4 completion: 'Explain quantum computing'"
Troubleshooting
"Invalid API key" Error
Verify your key starts with sk_live_
Check for extra spaces or quotes
Regenerate key at orgo.ai
Ensure the header is properly formatted
For mcp-remote: use Key:Value format (no space after colon)
Check environment variable is set
"Connection refused" Error
Cloud: Check https://orgo-mcp.onrender.com/health
Local: Ensure server is running on correct port
Check firewall/proxy settings
Claude Desktop Not Connecting
Restart Claude Desktop after config changes
Check config file syntax (valid JSON)
View logs: Help > Troubleshooting > Open Logs
Tools Not Appearing
Wait 10-30 seconds after connection
Check server logs for errors
Verify MCP server is listed in Claude settings
Architecture
Cloud Deployment:
Client (Claude) --> HTTPS --> Cloud Server --> Orgo API
X-Orgo-API-Key header
Local Deployment:
Client (Claude) --> stdio --> orgo_mcp.py --> Orgo API
ORGO_API_KEY env var
Project Structure
orgo-mcp/
├── orgo_mcp.py # MCP server (34 tools, dual transport)
├── pyproject.toml # Package configuration
├── Dockerfile # Production container
├── docker-compose.yml # Local development
├── render.yaml # Render.com deployment
├── fly.toml # Fly.io deployment
├── .dockerignore # Docker build exclusions
├── .env.example # Environment template
├── README.md
└── LICENSE
Development
# Install dev dependencies
pip install -e ".[dev]"
# Format code
black orgo_mcp.py
# Lint
ruff check orgo_mcp.py
# Run tests
pytest
# Test HTTP transport locally
MCP_TRANSPORT=http python orgo_mcp.py
# Test health endpoint
curl http://localhost:8000/health
# Test MCP endpoint
curl -X POST http://localhost:8000/mcp \
-H "Content-Type: application/json" \
-H "X-Orgo-API-Key: your_key" \
-d '{"jsonrpc":"2.0","method":"tools/list","id":1}'
License
MIT License - see LICENSE
Credits
