Installation

Usejarvis can run:

In Docker (recommended)
On your local machine via Bun
On a home server
On a VPS

Security First

Usejarvis can control files, terminals, browsers, remote channels, and connected machines. Treat installation as a security decision, not just a package install.

Before exposing the dashboard outside localhost or giving Usejarvis access to real accounts, read:

The daemon is the brain. It does not need to live on the same machine that you want to control. If you want Usejarvis to stay online 24/7, a server or always-on machine is the best fit. If you want Usejarvis to control a machine’s desktop, browser, or filesystem, install a sidecar there after the daemon is up.

Since 0.9, access is JWT-only by default. There is no shared password. You enroll each device with jarvis enroll, which mints a token the sidecar uses to connect. See Quick Start for the full flow.

Recommended Install: Docker

Docker runs Usejarvis on any OS with a single command. No Bun or other dependencies required.

1. Install Docker Desktop

Download Docker Desktop for Windows, macOS, or Linux if you do not have Docker yet.

2. Run the daemon

docker run -d --name jarvis \
  -p 3142:3142 \
  -v jarvis-data:/data \
  ghcr.io/vierisid/jarvis:latest

The image is published to GHCR for amd64 and arm64. All daemon state lives in the jarvis-data volume, mounted at /data.

LLM providers, API keys, and model routing are configured from the settings dashboard after first boot. They are stored in the database and encrypted keychain, not in env vars or config.yaml. Non-LLM configuration can be provided via environment variables or by mounting a config.yaml into the /data volume.

Important limitation:

The daemon inside Docker cannot access your host desktop, browser, clipboard, or native apps.
To give Usejarvis desktop awareness and automation on a machine, install the sidecar on that machine. This is a required step, not an optional one. See Quick Start.

3. Enroll your first device

Because the daemon runs inside a container, run the enroll command through docker exec:

docker exec jarvis jarvis enroll "my-desktop"

This prints an enrollment token. You will paste it into the sidecar in the next step. Continue with Quick Start.

Other Installation Options

Option 1: Bun Package

Use this if you want a native local install and already have Bun:

bun install -g @usejarvis/brain
jarvis start

The first time you run jarvis start, the daemon boots in setup mode and the dashboard at http://localhost:3142 guides you through LLM provider, voice, and profile setup.

Option 2: One-Liner Installer

Use this if you want the repository installed under ~/.jarvis/daemon automatically:

curl -fsSL https://raw.githubusercontent.com/vierisid/jarvis/main/install.sh -o /tmp/jarvis-install.sh
less /tmp/jarvis-install.sh
bash /tmp/jarvis-install.sh
jarvis start

If you want a more reproducible install, pin that URL to a specific release tag or commit instead of main.

Option 3: Manual Repository Install

Use this if you want the full source checkout:

git clone https://github.com/vierisid/jarvis.git ~/.jarvis/daemon
cd ~/.jarvis/daemon
bun install
bun run build:ui
bun link
jarvis start

Requirements

Before installing, make sure you have:

Docker, or Bun for a native install
A supported OS: Linux, macOS, or Windows via Docker or WSL2 (the daemon does not run on native Windows; the sidecar does)
An API key for at least one LLM provider (configured in the dashboard after first boot)

Optional but common:

Ollama if you want local models
Google OAuth credentials for Gmail or Calendar features
Telegram or Discord bot credentials for remote messaging
ElevenLabs credentials if you want premium TTS

Docker: Mounting a Real Config File

If you want repeatable management of non-LLM config, mount your own config.yaml into the container instead of relying only on environment variables.

Example pattern:

mkdir -p ~/.jarvis-docker
cp ~/.jarvis/config.yaml ~/.jarvis-docker/config.yaml

Then run:

docker run -d --name jarvis \
  -p 3142:3142 \
  -v ~/.jarvis-docker/config.yaml:/data/config.yaml \
  -v jarvis-data:/data \
  ghcr.io/vierisid/jarvis:latest

Remember: LLM provider keys are managed in the dashboard, not in this file.

Docker: Networking Advice

When Usejarvis runs in Docker, localhost inside the container is the container itself.

That means:

ollama.base_url: http://localhost:11434 only works if Ollama is inside the same network context the daemon can actually reach
the same warning applies to local Whisper endpoints and similar local services

If those services live somewhere else, point Usejarvis at:

a reachable host IP
a service/container hostname
a reverse-proxied hostname

See Troubleshooting for the detailed explanation.

Choosing Where to Run the Daemon

Local Machine

Best when:

You are just getting started
You want the simplest setup
You mainly use one machine

Tradeoffs:

Usejarvis only stays online while that machine is on
If you want it to start automatically after reboot, configure Autostart

VPS or Home Server

Best when:

You want Usejarvis available 24/7
You want one daemon to coordinate multiple machines
You care about uptime more than local-only simplicity

Tradeoffs:

You need to set daemon.brain_domain with an explicit scheme (http:// or https://) so sidecars dial the right endpoint
Desktop/browser control still happens through sidecars on the target machines

After Installation

The next steps are:

Run Quick Start to enroll a device, connect the sidecar, and finish setup in the dashboard
Configure Autostart if you run natively and want the daemon managed as a background service
Install sidecars on any machines you want Usejarvis to control directly

Sidecar Reminder

The daemon and the sidecar are different things:

The daemon is the brain
The sidecar is the machine-level actuator

If you run Usejarvis in Docker or on a VPS and want it to see and control your laptop, install a sidecar on the laptop. See Desktop Control.