Getting a Gemma 4 model serving on a Cloud TPU involves a surprising amount of ceremony: finding a zone that actually has v6e capacity, requesting flex-start VMs, sizing boot disks, wiring Hugging Face tokens through Secret Manager, picking the right vLLM flags for TPU, and remembering to tear everything down before the billing meter embarrasses you.
I packaged all of that into tpu-management — a Claude Code skill plus an MCP server (tpu-devops) — so you can just ask Claude to do it.
What it does
The skill teaches Claude the full TPU serving lifecycle, and the MCP server gives it ~40 tools to execute it:
Find capacity: sweep every zone with quota for an available TPU (find_tpu), check dimensioned quotas (get_zones_with_available_quota), estimate cost before creating anything.
