Moss

Self-hosted · Alpha

An AI assistant for your everyday life, running on your own hardware.

Moss brings your notes, tasks, calendar, email, and personal context into one place. Talk to it to plan your day, find something you saved, keep track of commitments, or work through a goal.

Install Moss View on GitHub

Moss is in active alpha. Features and installation details are still changing, so expect rough edges.

Moss Today screen with work priorities, personal tasks, calendar, and wellness for a sample user

What you can do with Moss

Try asking: "What should I focus on today?", "Find my notes about the kitchen renovation," or "Create a task to follow up on this next week."

Your assistant, your providers, your data

Moss stores its database and files on your hardware. Data is private by default and sharing is explicit. Administrators cannot read other people's private data, and connector and AI credentials are encrypted at rest.

You bring your own AI access. Moss works with Claude, Codex, and Gemini sign-in, plus configurable providers for features that need them. Your provider accounts and usage are separate from Moss.

Self-hosted storage does not mean all AI processing is local. When you use a hosted provider, prompts and relevant context go to that provider, and connected services and web tools make their own outside requests.

A look inside

Sample data for a made-up user, Luke, who mixes research work with everyday life.

Moss Tasks with Work, Personal, and Home lists
Tasks. Work, personal, and home lists side by side, with priorities and due dates.
Moss News front page with a lead story and source headlines
News. A reading desk for your sources, with summaries and links to the original reporting.
Moss Sports with followed teams, scores, and upcoming games
Sports. Results, upcoming games, and coverage for the teams you follow.
Moss Wellness with a mood check-in and recent trends
Wellness. Daily check-ins and medication tracking, with patterns over time.

Install with Docker Compose

You need Docker Engine or Docker Desktop, Docker Compose v2.24 or newer, and a terminal with curl. The commands use a POSIX shell, so Linux, macOS, or WSL. You do not need Node.js or the source code.

1. Download the deployment file

mkdir moss
cd moss
curl -fL https://raw.githubusercontent.com/motioneso/moss/main/infra/docker-compose.prod.yml \
  -o docker-compose.prod.yml

2. Generate your configuration and secrets

JARVIS_IMAGE_TAG=stable POSTGRES_PASSWORD=setup JARVIS_CLI_RUNNER_RPC_SECRET=setup \
  docker compose -p moss -f docker-compose.prod.yml --profile setup \
  run --rm --no-deps --pull always --user "$(id -u):$(id -g)" setup

This writes env.production.local next to the Compose file. Back it up somewhere safe, keep it out of Git, and run setup only once.

Before starting, add this line to that file, using the email you will sign up with:

[email protected]

3. Start Moss

docker compose -p moss -f docker-compose.prod.yml --env-file env.production.local \
  up -d --no-build

Open http://localhost:1533, create your first account with the email above, and follow the setup wizard to connect your AI provider. The first start takes a few minutes while the database and modules set themselves up.

Need remote access, HTTPS, backups, or upgrade steps? See the deployment guide, the HTTPS guide, and the release notes.