Labsco
MCP SERVER

RunPod MCP Server

by antonioevans

Run the RunPod REST API from a chat: pods, serverless endpoints, templates, network volumes and registry auth.

Cloud Resources & Infrastructure as Code
Summary
Full write access to your GPU spend, in a chat window.

Being able to say "spin up a 4090 with this image" and have the pod exist is genuinely faster than the console, and the same coverage extends to endpoints, templates and volumes. It is also a delete surface: everything that can be created here can be destroyed here, so this is a good place for a scoped key and a habit of reading back what it is about to do.

What it is

A server over the RunPod REST API that covers the whole lifecycle of the resources you pay for. GPU pods, serverless endpoints, templates, network volumes and container registry credentials are all managed from the client rather than the console.

What you get
  • Pods: create, list, get details, update, start, stop and delete
  • Serverless endpoints: create, list, get details, update and delete
  • Templates: create, list, get details, update and delete
  • Network volumes: create, list, get details, update and delete
  • Container registry authentications: create, list, get details and delete
Requirements

Node.js 18 or higher and a RunPod API key from the RunPod console, set as `RUNPOD_API_KEY`. Install through Smithery with `npx -y @smithery/cli install @runpod/runpod-mcp-ts --client claude`, or clone and `npm run build`. The key grants full access to your RunPod account and this server can start and delete paid resources with it — issue a separate, limited key rather than reusing your main one. MIT licensed.

Setup effort

One command plus a key — npx -y @smithery/cli install @runpod/runpod-mcp-ts --client claude, then supply credentials