Stats
Actions
Tags
From layerbrain
Use when the user wants to create, list, manage, snapshot, restore, SSH into, or check compute regions for Layerbrain machines (VMs / sandboxes / GPU instances). Requires the `brain` CLI.
How this skill is triggered — by the user, by Claude, or both
Slash command
/layerbrain:brain-machinesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Layerbrain machines are VMs / sandboxes / GPU instances. This skill covers the full machine lifecycle, snapshotting, and the region-availability check you do **before** creating one.
Layerbrain machines are VMs / sandboxes / GPU instances. This skill covers the full machine lifecycle, snapshotting, and the region-availability check you do before creating one.
Run brain whoami. If it errors, tell the user to run brain login (or set LAYERBRAIN_API_KEY). See references/auth.md.
machines create)| Intent | Command |
|---|---|
| List regions + available compute | brain compute list --output json |
| Get one region's details | brain compute get <REGION_ID> --output json |
Pick a region with the resources the user needs. Pass its ID to --region on machines create.
| Intent | Command |
|---|---|
| List machines | brain machines list --limit 50 --output json |
| Get a machine | brain machines get <ID> --output json |
| Create a machine | brain machines create --name <n> --image <img> --cpu <int> --ram-gb <num> --disk-gb <num> --region <REGION_ID> --ttl-minutes <int> --output json |
| Create with GPU | add --gpu <type> (e.g. --gpu a100) |
| Attach to networks | add --networks <id1,id2> |
| Extend TTL | brain machines extend <ID> --duration-minutes <int> (use -1 for perpetual) |
| Restore (after stop) | brain machines restore <ID> |
| Delete | brain machines delete <ID> |
| SSH | brain machines ssh <ID> --user root |
For complex create payloads, use --data-file ./machine.json instead of repeating flags.
Snapshots are machine-scoped. Use them to checkpoint state before risky ops or to clone.
| Intent | Command |
|---|---|
| Snapshot a machine | brain machines snapshot <ID> --name <label> --output json |
| List snapshots | brain snapshots list --output json |
| Get a snapshot | brain snapshots get <ID> --output json |
| Restore from snapshot | brain snapshots restore <ID> |
| Get download URL | brain snapshots download <ID> |
| Create raw snapshot | brain snapshots create --data-file ./snap.json --output json |
brain compute list before brain machines create unless the user already specified a region.id from JSON after create or snapshot — you'll need it for follow-ups.--ttl-minutes matters: machines without TTL extension auto-expire. Warn the user if creating without --ttl-minutes for long-running workloads. To make a machine perpetual, pass --duration-minutes -1 to extend.delete is destructive. Suggest snapshot first if the machine has state worth keeping.brain machines ssh attaches a TTY.Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub layerbrain/skills --plugin layerbrain