maze-labs

Tools for homelabs, from two people who run one.

We're two homelab enthusiasts. We kept finding software built for companies, or for a single box under the desk, and wanted tools that fit a homelab and leave us free to run things our way. So we started writing them.

~/homelabzsh

~/homelab $ ls maze-labs/

ipmaze/● coming soonip address management
mazeamt/● coming soonintel vpro, out of band
mazecd/● availableargo cd on your phone
mazedns/● availablemulti-site dns filtering
mazeregistryui/● availablecontainer registry browser

route.trace0 steps

~/maze-labs/ipmaze● coming soon

Flagship

IPMaze

IP address management for homelabs and small networks. Sites, networks, Wi-Fi, hosts and the VPNs between them, documented in one place instead of a spreadsheet.

Sites, networks, SSIDs, hosts, interfaces, VIPs, clusters and VPNs
Sites on a map, with a speedtest agent for bandwidth and public IP
REST API with interactive docs and tokens for your scripts
Local accounts with TOTP 2FA, or single sign-on via OIDC
Audit log and JSON export of the whole workspace
SQLite by default, PostgreSQL when you need it

// self-hosted today. a SaaS edition, hosted on our own infrastructure, is a work in progress.

IPMaze network view with an address map and host list

team

2 people · 1 homelab each

~/maze-labs/team/AndreaF17

Andrea Ferrario

@AndreaF17 · Varese, Italy

github linkedin

~/maze-labs/team/M4RC02U1F4A4

Marco Galvagni

@M4RC02U1F4A4 · Italy

github linkedin

projects

3 available · 1 coming soon

~/maze-labs/mazedns● available

MazeDNS

A filtering DNS resolver in the spirit of Pi-hole and AdGuard Home, built for more than one site.

One control plane for config and dashboard, DNS agents close to your clients
Agents enroll with a key and keep resolving if the control plane goes down
Blocklists, allow and deny rules, rewrites, local zones and split-horizon
DoH, DoT, DNSSEC, OIDC single sign-on and Prometheus metrics

get started github

~/maze-labs/mazeregistryui● available

MazeRegistryUI

A server-rendered web UI for your OCI and Docker registries, light enough for a phone.

Several registries behind one picker, each with its own credentials
Layer composition, multi-arch indexes, labels and the raw manifest
One binary and one config file: no Node, no CDN, runs air-gapped
Credentials stay on the server; the browser never talks to the registry

get started github

~/maze-labs/mazecd● available

MazeCD

A mobile-first PWA that watches all your Argo CD instances from one screen.

Install it on your phone; it updates itself seconds after a new release
Every application from every instance, with resource tree, pod logs and git diff
The Argo CD token decides: read-only, or sync and resource actions too
One Go binary proxies to Argo CD, so tokens never reach your device

get started github

~/maze-labs/mazeamt● coming soon

MazeAMT

A web UI for out-of-band management of Intel AMT (vPro) PCs, remote desktop included.

KVM in the browser, through an RFB client written for the AMT firmware
Power control and one-time boot to PXE, disk, CD/DVD or BIOS setup
Hardware inventory and firmware event log, with no agent on the machines
Built for a fleet: one dashboard with power state, tags and search

// source not public yet

~/maze-labs/BUILT_WITH_AI.md● ai-assisted

We build with AI, and we say so.

We're not professional software developers. We're IT graduates with a development background, but by day one of us works in cybersecurity and the other as a DevOps / platform engineer. That's where our sense of what a homelab really needs comes from.

We write these tools together with AI coding assistants, and we'd rather tell you upfront than have you find out from the commit history. AI helps us move faster, but it doesn't make the decisions.

We review every change, run everything in our own homelabs before it's released, and read the code with the same security mindset we bring to our work.

If something doesn't look right, please open an issue. We'd rather hear about it than have it go unnoticed.

we designWe decide what each tool does, how it behaves and what it deliberately leaves out.
we testEvery change runs in our own lab before it ships. If it breaks, we find out first.
you checkPublished projects are open source, so you can read every line yourself.