Sparehand

Docs

Short guides, each with the exact commands. Start with Getting started. Features marked Beta work and are included, but may still change.

Getting started

The first ten minutes with the Mac app: opening a build macOS can't verify yet, Setup's checklist, pairing your first machine, your first sparehand run, and where to send a problem.

Requirements

What the Mac needs (macOS 14, Apple silicon or Intel), what each kind of spare machine needs, and what the beta features ask for on top: an NVIDIA card, disk space, API keys, agent accounts, Tailscale.

Support

Where to write, what to put in the message (the app's Copy Diagnostics, the logs), and how to report a security problem privately.

Adding a machine

Step by step for a Mac, a Linux machine or a Windows PC: set up the connection, check the machine is really yours, and lock it down so only your Mac can use it.

Settings

Where Sparehand keeps its settings, how to change limits such as memory and CPU, add setup steps for a project, and what exactly happens during a run.

AI coding assistants

Let Claude Code, Gemini CLI, Cursor, Codex and other coding tools send their tests, type checks and lint to spare machines automatically. Explains what is handed over (checks), what stays on your Mac (builds, installs, git), and which tools connect by a hook and which by a note.

Coding tools' hooks

For each coding tool: the file Sparehand writes, what the tool hands its hook and what the hook answers, with the page each was checked against and when.

Travel mode

Away from home with a laptop: keep working on your Mac and its spare machines, with your code, the jobs and the machines all staying at home.

Local AI models

Install the Ollama model server on a spare machine from your Mac, download models, and use one address, localhost:11434, for all of them.

Graphics card jobs

Send AI, speech-to-text or training work to a machine with an NVIDIA graphics card, one job at a time, without mistaking it for someone gaming.

Waking and sleeping

How sleeping machines are woken when needed, kept awake during long jobs, and optionally put to sleep when idle.

Reaching a machine's services

Make a model server, database or dev server on a spare machine answer on your Mac, through a private tunnel only your Mac can use.

GitHub Actions on your own machines

Make a spare machine a self-hosted runner for a repository in one step, with each job counted as one of its jobs.

Android emulators

Boot Android emulators on your spare machines instead of your Mac, and reach each one from adb here with a serial of its own.

Mobile builds

Xcode, Gradle, Flutter and React Native builds on a spare machine that has the toolchain. The second build is incremental, and the app, the APK and the test reports come back.

File sharing

Show a folder from a spare machine in your Finder, set up from your Mac. Only your Mac can connect.

Using it from anywhere

Open the dashboard on your phone or tablet away from home through Tailscale, without opening anything on your router.

Screens

View and control a machine's screen from the dashboard, handy for checking on a spare computer from your tablet.

Work items

Optional: list your Jira or Azure DevOps tasks in the dashboard and start a plan-first Claude Code session for one.

Security model

Who trusts whom, what leaves your Mac, why .env files stay home, and how to lock down each machine.

Uninstall

Everything Sparehand puts on your Mac and on each machine, and how to take all of it off: sparehand remove NAME --clean, the Keychain items, the folders.

Names, words and colours

The names Sparehand uses (this Mac, workers, jobs), the one word for each state a machine can be in and the colour that goes with it, for anyone building on Sparehand or translating it.

Changelog

What each version added, changed or fixed.

Roadmap

Ideas under consideration, such as smarter placement, shared caches and notifications.

Beta Beta

These features are part of the app and the licence. They work, but they are newer, lean on outside services or accounts you bring, and may change between versions. Each page says what it needs.

Sessions

Your Claude Code sessions on the hub, followed and driven from any device: answer prompts, see each turn's changes, run checks on a spare machine, get notified on your phone.

Machines in the cloud

A Runpod, AWS, Azure or Google Cloud machine as a worker, with a GPU or without, started when a job needs it and stopped when idle, with caps on hours and spend.

Crew

Give one task to a Master AI; it splits the work between agents from different providers, reviews each part and hands you one branch.

Studio

Describe a video; a Director plans the storyboard, picks the tools (voices, pictures, clips, stock footage) and hands you the finished files, in every language you ask for.

The Advisor

Lessons kept from Crew's and Studio's own corrections, and recommendations drawn from what Sparehand notices about your setup.

Members and credits

Let friends use your AI models, Studio and Crew on prepaid credits: a People tab, prices by model size, a ledger per person, API keys for their own apps.