Skip to main content

Usage

Description

Pull an MCP server from the R2 registry and configure it for your AI client. The config format differs by client:
  • VS Code: native HTTP transport (type: http) - no proxy needed
  • All other clients: stdio bridge via npx mcp-remote

Options

string
required
MCP server name to pull from registry
string
required
Target AI client: vscode, cursor, antigravity, claude, or chatgpt

Supported Clients

VS Code

Configures VS Code MCP settings

Cursor

Configures Cursor MCP settings

Antigravity

Configures Antigravity MCP settings

Claude

Configures Claude Desktop MCP

ChatGPT

Configures ChatGPT MCP settings

How It Works

1

Fetch from R2

Retrieves server metadata from R2 registry (entrypoint, lang, repo URL)
2

Generate Config

Creates client-specific MCP configuration with the Worker HTTP URL
3

Write config entry

  • VS Code: writes { "type": "http", "url": "..."} under the "servers" key in .vscode/mcp.json
  • Other clients: writes { "type": "stdio", "command": "npx", "args": ["-y", "mcp-remote", url] } under the "mcpServers" key in the client’s config file
4

Save Config

Writes configuration to the client’s config file location

Architecture

Examples

Example Output

Configuration Format

The generated configuration depends on your client.
Written to .vscode/mcp.json under the "servers" key:
VS Code supports the HTTP transport natively - no proxy process required.

Platform Support

Next Steps

Inspect Server

View server details

Test Mode

Test without registry