Remote Compute: Modal
SkillProductivityThis skill sends your Feynman research notebooks and experiment jobs to Modal, a cloud computing service. Once added, your AI can run research workloads remotely instead of only on the local machine, including jobs that need GPUs. Runs are bounded and reproducible, so results stay consistent.
Available today. Use it from your connected AI after setup.
No other account needed.
After adding it, pick a notebook or experiment job and ask your AI to send it to Modal. Use it when a task has specifically chosen Modal for its compute.
Then ask your AI: use the Remote Compute: Modal skill
What your AI can do with it
- Run Feynman research notebooks on Modal's cloud
- Dispatch experiment jobs to Modal for remote execution
- Handle GPU tasks that a local machine can't run
- Keep cloud compute within set bounds
- Repeat runs with reproducible results
What this skill tells your AI
The instructions your AI receives, as published by companion-inc/feynman in skills/remote-compute-modal/SKILL.md and read by ahel’s review.
Use this skill only after Modal is the chosen execution route.
Workflow:
- Verify Modal credentials and CLI availability through Feynman Settings or a non-secret CLI check.
- Keep the job bounded: inputs, package installs, hardware, timeout, and expected artifacts must be explicit.
- Submit through Feynman's notebook Modal path or a recorded Modal script.
- Save remote URL/handle, command, stdout/stderr, returned artifacts, and environment snapshot.
- Stop, retry, or mark failed jobs through the workbench compute lifecycle instead of hiding failures.
Do not use Modal for unrelated hosting or indefinite services from this skill.
Signals
- GitHub stars
- 9k
- Forks
- 1k
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
remote-compute-modal- Source
- github.com/companion-inc/feynman