remote-gpu

Orchestrate headless GPU workloads via SSH or HTTP API.

3|2|Updated Mar 17, 2026

Search Tags:#automation #orchestration #gpu #ssh #claude-code #compute

Ready to run

npx skills add https://github.com/broomva/skills --skill remote-gpu

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: remote-gpu
Source: https://github.com/broomva/skills/tree/main/skills/compute/remote-gpu
Command: npx skills add https://github.com/broomva/skills --skill remote-gpu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires fastapi, uvicorn, psutil, httpx, rsync, jq, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill solves the friction of managing headless GPU hardware by providing a unified interface to submit, monitor, and retrieve results from remote servers without manual SSH session management.

Core Features & Use Cases

Job Orchestration: Submit training, inference, or agentic workflows to remote GPU hardware with automated queue management.
Remote Development: Launch interactive Claude Code sessions directly on high-compute remote machines.
Monitoring & Sync: Real-time GPU telemetry and automated file synchronization between local workstations and remote servers.

Quick Start

Use the remote-gpu skill to submit a training job to the server by running the gpu-submit command followed by your training script and arguments.

Frequently Asked Questions about remote-gpu

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I orchestrate remote GPU training jobs from my local machine?▼

You can orchestrate remote GPU training jobs by submitting scripts via SSH or an HTTP API. The system manages the job queue, streams logs, and synchronizes files between your local workstation and the headless server automatically.

Can I run interactive Claude Code sessions on a headless GPU server?▼

Yes, you can launch interactive Claude Code sessions directly on remote high-compute machines. This allows you to leverage remote GPU hardware for development and agentic loops without manual SSH session management.

What do I need to set up before submitting inference workloads to remote GPU hardware?▼

Before submitting inference workloads, you need SSH access to the remote server, rsync installed locally, and Python-based server components. These prerequisites enable command execution and file synchronization across distributed compute environments.

How does file synchronization work between my Mac and distributed compute environments?▼

File synchronization relies on rsync to transfer data between your local workstation and remote servers. It operates automatically alongside job orchestration to ensure your training scripts and output files remain consistent across environments.

Does this approach to GPU orchestration support real-time hardware monitoring?▼

Yes, the orchestration approach supports real-time GPU telemetry. It uses Python server components to monitor remote hardware status, allowing you to track active workloads and system health during training or inference jobs.

What are the limitations of using FastAPI and SSH for headless server orchestration?▼

Using FastAPI and SSH for headless server orchestration requires continuous network connectivity and manual setup of Python server components on each remote machine. It does not auto-provision hardware, meaning pre-configured SSH access is mandatory for job submission.