Consistently changing 10% of the aquarium water weekly is essential to this system. A 10% swap exports accumulated organics, replenishes trace elements that no test kit measures, and dilutes whatever is slowly drifting in the wrong direction before it becomes a problem worth diagnosing. It's the cheapest insurance in the room.
Automatic wins. It never gets skipped, never doubles up, and the tank never swings. Manual is big moves less often — larger impact, can cause swings, easy to forget.
WaterLogged is a single-file HTML aquarium tracker built with Claude. It is designed to track parameters, livestock, health issues, protocols, and maintenance across multiple tanks — saltwater and freshwater — over time. In its simplest form you could use it for years by just opening a single file in any web browser, with no server, no database, and nothing to install.
If you are willing to subscribe to Claude, the power starts to really show. At its most basic, Claude can monitor your data and help plan daily tasks to keep you on track as an aquarist. Even at this level the capability is remarkable. As trust in AI grows, Claude can evaluate both specific measurements and anecdotal observations, compare photos over time to spot differences, and identify trends and patterns in your aquarium. Ultimately — suggesting corrective actions before problems become losses.
Version: v4 · Storage key: waterlogged-data · Built May 2026
After watching a Serious Reefs video featuring Claude, it quickly became clear that AI could be something more than a novelty — it could actually help manage an aquarium. In a very short time, partly at Claude's own suggestion, it felt right to build a tracker for the tanks. No programming background required, and none was needed.
The code behind WaterLogged is a mystery to its creator, and that's part of the point. What matters is what it does: the database it builds over time — parameters, observations, livestock, trends — is more complete and more useful than anything that could have been assembled from lists, photos, or testing alone. It just keeps getting better.
AI is not a popular idea in the online or local reefing community. People say it makes mistakes, that it doesn't always get it right. In a relatively short time using it — yes, it has made mistakes. But in that same time, so have humans. When people say AI gets things wrong, you have to stop and think about where AI learns everything it knows. The truth is that humans don't have the correct answers either. Online forums, product manufacturers, local fish store owners, and fellow aquarists are still debating what works and what doesn't. That's not AI's fault — that's the state of the hobby.
Why are we still debating? Because for most problems we try two, three, five, ten different things over a relatively short period with no control group — and in reality we never know which step actually solved the problem. It's too hard for one person, even with several tanks, to answer definitively.
But what if there's a tool that can process enormous amounts of data and find patterns across thousands of tanks and thousands of issues? AI doesn't learn the way humans learn — it looks for patterns and figures out how things work. If the reefing community started feeding data into a shared system at scale, could AI start finding real answers? That is exactly what AI was built to do.
WaterLogged is probably not the program that will ultimately do that. But it might be the one that opens people up to trusting AI — and to sharing the vast amounts of data that AI needs to find those answers.
This app lives in two Claude chats. The Master TankTracker chat builds and develops the app — new features, UI, architecture. The Journalling my aquatic life chat handles daily tank observations, parameter logging, and Sunday data rebuilds into the app.
Historical data is hardcoded in the app and rebuilt each Sunday. New entries during the week save to window.storage and merge over the hardcoded defaults on load. The JSON export/import system (coming soon) will make data fully portable — usable in Claude, standalone in a browser, or hosted on Netlify.
WaterLogged is designed to eventually run without Claude at all. One HTML file, opened in any browser, with data saved locally. No subscription, no server. Share it with a friend and they have a full aquarium tracker from day one.
Exports all parameter records, observations, coral and fish health issues, maintenance entries, and full inventory to a single JSON file. Store it anywhere — Google Drive, Dropbox, your desktop. Use it to move data between devices or as a backup before major changes.
Select a WaterLogged JSON export file to load. The app will show you a summary of what's in the file before anything is changed. Importing will replace all current data — export first if you want to keep it.