MOSS
B1 · IT DEPT

Open source · runs on your own hardware

Your AI IT department,
in your basement.

MOSS looks after your home or small-office network with a team of AI agents. They find every device, watch the services you care about, work the problem when something breaks, and suggest fixes. Nothing changes without your say-so.

Get started View on GitHub
The Basement: MOSS's agents at their desks. Nina is working an incident and Iris a security check, while the others are on a break.

What MOSS does

Hire a team. Keep the keys.

You hire agents (a Network Admin, a Systems Admin, a Security Admin) and give them a budget. They do the watching and the digging, so you only step in when it matters.

01

Knows what's on your network

Agents scan the networks you allow and keep an inventory of every device: what it is, its address, who made it. Lock one and they leave it alone.

02

Watches your services

Ping, web, port, certificate and DNS checks, or alerts from Uptime Kuma, Beszel and Alertmanager. Build dashboards and put one on a TV.

03

Works the problem

When something goes down, MOSS opens an incident and an agent starts investigating straight away, then tells you what it found.

04

Fixes things, with your approval

Restarting a service, rebooting a server, blocking a device: an agent writes down the exact steps and how to undo them, then waits for you.

05

Agents you can talk to

Chat with any agent, give it a task, or ask Moss, who knows MOSS inside out. Each one has its own skills, schedule and spending limit.

06

Any AI model, even a local one

Anthropic, OpenAI, OpenRouter, a Claude Pro or Max subscription, or Ollama on your own hardware, so nothing leaves your network.

How it keeps you safe

The AI can be wrong. MOSS plans for it.

Your network can contain hostile data: a device named "ignore previous instructions" is a real attack. So safety never depends on the AI behaving.

  • No shell. Agents only use specific, checked tools. None runs arbitrary commands.
  • A gate checks every action against the kill switch, the agent's budget and permissions, and the networks you allow.
  • Changes need approval, step for step, from MOSS or from your phone.
  • Secrets never reach the AI. Agents only see a name like secret:router-ssh.
  • Everything is recorded in an audit log you can verify hasn't been altered.
CHANGE REQUEST · CR-12Waiting for approval

Restart Plex on the media server

Plex stopped answering at 21:04. The container is running but unhealthy after a failed library scan.

  1. container_restart plex on 192.168.50.20
  2. http_check 192.168.50.20:32400

Rollback: none needed. The restart changes no settings.

ApproveReject

A look around

Everything in one place

Dashboard: Moss's briefing, counts of open incidents, monitors down, changes awaiting approval, assets, agents and spend
The dashboard: what your IT department is doing, at a glance.
Monitoring: monitors with their state, last result and responder agent
Monitoring, with an agent ready to respond to each check.
Assets: every device on the network with its kind, address, MAC and maker
Every device on your network, and what agents may do with it.
An agent's page, with suggested tasks, recent runs, skills and its mascot
Each agent: give it a task, see its work, pick its look.
Settings, Integrations: each integration with a switch to turn it on

Integrations

Works with what you already run

Each integration adds health checks, fills in your inventory, and gives agents safe, approved ways to help.

Home & home lab

Free
  • Home Assistant
  • Proxmox
  • Docker
  • Pi-hole
  • AdGuard Home
  • OPNsense
  • pfSense
  • UPS (NUT)
  • Tailscale
  • Cloudflare
  • Telegram
  • Discord

Business

Licence
  • Microsoft sign-in
  • Google sign-in
  • Microsoft 365
  • Google Workspace
  • Slack
  • PagerDuty
  • FortiGate
  • Cisco Meraki
  • AWS

See every integration and what it does →

Get started

Up and running in minutes

You need a computer that stays on (Linux is recommended), Docker with the Compose plugin, and openssl.

$ git clone https://github.com/Dan5672/MOSS.git
$ cd MOSS
$ git checkout "$(git tag --list 'v*' --sort=-v:refname | head -n 1)"
$ sh deploy/init.sh
$ docker compose -f deploy/docker-compose.yml --env-file deploy/.env up -d --build
  1. Open MOSS at the address init.sh printed, from any device at home, and create your account.
  2. Allow your network so agents know where they may work.
  3. Add an AI model: an API key, a Claude subscription, or a local Ollama.
  4. Hire a Network Admin and let it discover your network.

New to MOSS? The getting started guide goes step by step.

Free & business

Free for your home. Ready for your business.

Community

Free

Open source under the GNU AGPL v3. Everything you need for a home or home lab.

  • Unlimited agents, devices and monitors
  • Every home and home lab integration
  • Approvals in Telegram or Discord
  • Runs entirely on your own hardware
Get started

Business

Licence

For companies that run on Microsoft, Google, Slack or the cloud.

  • Sign in with Microsoft or Google
  • Microsoft 365 and Google Workspace
  • Slack and PagerDuty
  • FortiGate, Cisco Meraki and AWS
Ask about a licence

Contact

Get in touch

Ask about a business licence, tell us how you use MOSS, or ask a question. We'll reply by email.

  • Found a bug or have an idea? Open an issue on GitHub.
  • A security problem? Please use this form, not a public issue, and choose "Security report".