Skip to content
Get packs

Modal Cloud GPU Runner

Run training, fine-tuning, and inference on rented cloud GPUs via Modal, with cost estimates first.

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
1 file

What it does

When installed, the agent can write Python launcher scripts that run your code on remote GPUs rented from Modal, a serverless cloud provider, instead of your own machine. Before every run it must estimate the cost and show it to you for confirmation, and it suggests a GPU based on the model size and VRAM needs. The skill also guides login, budget and spending-limit setup, and how to package dependencies so runs are reproducible.

Great for

  • Developers without a local GPU who need to debug CUDA code or run small experiments
  • Researchers who want to launch a training or benchmark job without managing servers or Docker
  • People who want pay-per-second GPU billing with no idle instances left running

Skip it if

  • Long training runs over several hours, where the skill itself suggests cheaper providers, because Modal costs more per GPU-hour for some tiers
  • Anyone who must avoid sending code or data to a third-party cloud, since the GPU function runs remotely on Modal's infrastructure

Try asking

  • “Run my fine-tuning script on an A100 on Modal and estimate the cost first”
  • “Set up a Modal inference web endpoint for my 7B model”
  • “Benchmark this model on H100 vs L4 on Modal and tell me the cost per 1000 samples”
Step-by-step guide: how to use Modal Cloud GPU RunnerInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    serverless-modal.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip serverless-modal.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip serverless-modal.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Set a workspace spending limit on modal.com before any run, as the skill recommends, and never enter card details through the agent.
  • Review the cost estimate the agent shows before confirming each run, since the skill requires it for every run.
  • Start with a short test job on a cheap GPU like T4 or L4 to check that the code works before launching a long job on an expensive GPU.

What's inside

  • SKILL.md

Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.