← Back to MCP Servers
████
█>>█
█>>█
████

On-Demand Compute Engine

v1.0.0ComputeTools 4 actions

SERVER URL

http://localhost:3000/api/mcp-server/compute-engine

CONNECTION JSON

{
  "mcpServers": {
    "compute-engine": {
      "command": "npx",
      "args": [
        "mcp-remote@latest",
        "http://localhost:3000/api/mcp-server/compute-engine"
      ]
    }
  }
}

Learn more about supported MCP clients

MCP HOST CONNECTIONS

Connect this MCP server directly to your AI assistant tools by using the JSON config above. Click on a logo to see a demo video.

╔═══╗
║ ▸ ║
║ │ ║
╚═══╝
CursorWatch demo
┌───┐
│ C │
│ L │
└───┘
ClaudeWatch demo
≋≋≋≋
~≋~≋
≋~≋~
~~~~
WindsurfWatch demo

DESCRIPTION

Serverless compute platform with x402 micropayments. Run Python, Node.js, Rust code on-demand. Train ML models, process large datasets, or run intensive computations. Pay only for execution time.

AVAILABLE TOOLS

x402_compute_credits

Check compute credits and payment status

execute_code

Execute code (Python/Node/Rust) with x402 payment

train_model

Train ML model on GPUs (charges via x402)

batch_process

Process large dataset (charges via x402)

CODE EXAMPLES

JavaScript
// Compute Engine with x402
import { Client } from '@modelcontextprotocol/sdk';

const client = new Client();
await client.connect('https://compute.x402.app/mcp');

// Execute code (auto-pays for compute time)
const result = await client.callTool({
  name: 'execute_code',
  arguments: {
    language: 'python',
    code: 'import pandas as pd\ndf.describe()',
    wallet: 'YourWallet...'
  }
});

Official MCP SDK Documentation:
https://github.com/modelcontextprotocol/modelcontextprotocol

Installation:
npm install @modelcontextprotocol/sdk

Authentication:
- For MCP Hosts (Cursor, Windsurf, Claude Desktop), saving the JSON config to your tool will automatically trigger the browser OAuth flow
- For programmatic access, obtain the OAuth token by running "npx mcp-remote <url>" in the terminal and follow the instructions

x4