Stop Forgetting What Your Containers Do!
Run `docker ps` on any server that’s been alive for more than a few months, and you’ll probably see it: a container with a name like `app_prod_2` or `worker-final-v2`, quietly running, doing… something. You set it up eighteen months ago. It was important then. Is it still? What does it actually do? How would you even begin to back it up if the drive died tonight?
If that scenario feels uncomfortably familiar, you’re not alone — and it’s exactly the problem Dockyard was built to solve.
The problem: Docker remembers what, never why
Docker is excellent at running containers. It is completely indifferent to whether anyone remembers what they’re for. A `docker inspect` will tell you the image, the ports, the mounted volumes — the facts. It will never tell you that this particular container is the DNS server for your entire network, or that the config file lives on a different drive than you’d expect, or which command actually restarts it safely.
That knowledge lives in your head, in a Notion page you stopped updating, or nowhere at all. And it evaporates the moment you forget, move on, or hand things off to someone else.
What Dockyard actually is
Dockyard is a free, open-source, self-hosted dashboard that sits alongside your existing containers and does one job well: it documents them, automatically where it can and by hand where it can’t.
For every container, it quietly merges three layers of information into one place:
– Live facts, pulled straight from the Docker API — status, ports, volumes, networks, labels — always current, no manual upkeep required.
– Compose file context, so you can see exactly how a service was declared, including `depends_on` relationships and anything not visible from a running container alone.
– Your own notes — a summary, usage instructions, debugging tips, backup steps, links to upstream docs. The part no automated tool can generate for you, because it’s the part that only you know.
The result is a single page per container that answers the only question that actually matters at 2am when something breaks: what is this, and what do I do about it?
A few things worth knowing about it
It knows when documentation is missing. New containers are flagged automatically, and anything without at least a summary gets a clear “needs docs” warning — so gaps in your knowledge base show up before you need them, not during an outage.
Nothing gets deleted by accident. When a documented container disappears, Dockyard archives it instead of quietly forgetting it ever existed — the notes stay put, in case you need them again.
It scales to more than one server. Point it at several hosts and get one dashboard for your whole fleet, with a lightweight agent for servers where you can’t share a filesystem directly.
It plays well with tools you already use. Traefik routing labels are read automatically, and if a container’s web UI is reachable, a one-click launch button opens it — smart enough to work out the right address even on more unusual network setups (macvlan, host networking) where a naive guess would send you to a dead page.
Your notes can travel with the container itself. A write-back feature can push key details into the container’s own Docker labels — so the same information shows up if you ever look the container up in Portainer or Podman, not just in Dockyard.
Who this is actually for
If you run a handful of containers on one machine and can hold it all in your head, you may not need this yet. But if any of this sounds familiar, it’s worth a look:
– If you’ve ever opened a terminal, stared at a container name, and had absolutely no idea what it does
– You manage more than one server and want a single view across all of them
– You’ve inherited someone else’s infrastructure and are reverse-engineering it container by container
– You’re the only person who knows how any of this works, and that fact keeps you up at night
Getting started
Dockyard is free, open source, and self-hosted — your data stays on your own infrastructure, always. It runs as a Docker container itself, deploys in minutes with a single `docker compose up`, and works immediately with zero configuration for a single-host setup.
Take a look…
