Runpod

Connects Claude to RunPod's GPU infrastructure API for spinning up compute on demand. You get tools to launch pods with specific GPU types, scale serverless endpoints across regions, monitor resource usage, and tear down instances when you're done. Useful when you need to orchestrate ML workloads from a conversation, like kicking off a training run on an A100, checking if your inference endpoint is warm, or programmatically managing a fleet of GPU workers without leaving your chat interface. Part of the MCP Armory collection, auto-generated from RunPod's OpenAPI spec and tested against their live API.