inbox.dog in Shelley

Shelley is an AI coding agent on exe.dev. It runs in a full Linux VM and supports MCP servers, so you can give it Gmail access via inbox.dog.

Official docs

Shelley documentation

Shelley supports MCP servers configured with environment variables. See the exe.dev docs for full details on MCP configuration.

1. Get credentials

Create an API key and connect Gmail at inbox.dog/connect.

2. Add MCP config

Add the inbox.dog MCP server to your Shelley configuration:

{
  "mcpServers": {
    "inbox-dog": {
      "command": "npx",
      "args": ["-y", "inbox.dog", "mcp"],
      "env": {
        "INBOX_DOG_CLIENT_ID": "your_client_id",
        "INBOX_DOG_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

Replace your_client_id and your_client_secret with the values from inbox.dog/connect.

Alternative: use the SDK directly

Since Shelley runs in a full Node.js environment, you can also use the inbox.dog npm package directly in a script:

import InboxDog from 'inbox.dog';

const client = new InboxDog({
  clientId: process.env.INBOX_DOG_CLIENT_ID,
  clientSecret: process.env.INBOX_DOG_CLIENT_SECRET,
});

const messages = await client.listMessages({ maxResults: 5 });
console.log(messages);

Install with npm install inbox.dog and set the environment variables in your session.

3. Start using it

The MCP server will be available in your next Shelley session. You can then ask Shelley to read, search, and send emails through Gmail.