tensorpool

This skill helps users run ML workloads on TensorPool GPU infrastructure. It supports two workflows — (1) Jobs, a fire-and-forget git-style interface for training experiments via tp job, and (2) Clusters, interactive SSH-based GPU development via tp ssh. Use this when a user wants to run scripts on cloud GPUs, submit training jobs, manage GPU clusters, or scale up to professional hardware (H100, H200, B200, B300, etc.).