Quickstart

Publish and use MCP servers with SuperBox

For developers: publishing a server

1. Install the CLI

   pip install superbox
   
2. Authenticate

   superbox auth login
   

   Follow the device flow instructions to authenticate via browser.

3. Initialize server metadata

   superbox init
   

   Provide your server name, repository URL, description, and entrypoint file.

4. Publish to the registry

   superbox push
   

   This runs the full security pipeline:

   • SonarCloud static analysis
   • MCP tool discovery
   • Snyk dependency vulnerability scan
   • GitGuardian secret detection
   • Bandit Python security linting
   • Upload metadata to the R2 registry

For users: connecting a server to your AI client

1. Browse available servers

   Visit superbox.1mindlabs.org/explore to find published servers.

2. Install the CLI

   pip install superbox
   
3. Pull the server config

   superbox pull --name weather-server --client cursor
   

   Supported clients: vscode, cursor, antigravity, claude, chatgpt.

   This writes the MCP configuration for your client. VS Code receives an HTTP-type entry pointing directly to the Cloudflare Worker. All other clients (Cursor, Antigravity, Claude, ChatGPT) receive an npx mcp-remote stdio bridge entry.

API usage

1. List all servers

   curl https://api.superbox.ai/api/v1/servers
   
2. Get server details

   curl https://api.superbox.ai/api/v1/servers/weather-server
   
3. Connect to your AI client

   superbox pull --name weather-server --client vscode
   

   The server appears in your AI client’s MCP tools list immediately.

Next steps

API reference

All endpoints

CLI commands

Full command reference

Authentication

Auth flow details

Error handling

Handle API errors

Architecture

System design

Security

Security pipeline