Waterllama

Promptfree, for everyone, and the only version there is

Copy the prompt for Claude CodeCopy the prompt for CodexCopy the prompt for Cursor

Copy

Build me a water-intake tracker in the shape Waterllama has: log a drink, see something happen, get a nudge when I am behind.

Read this first, because it is unusually honest for a technical prompt: the product being replaced costs a few euros once, and what you are paying for is illustration and animation — a character that reacts, drinks that look like drinks. That is design work, not engineering, and if you cannot do it or commission it, the version you build will be a counter with a progress bar. A counter with a progress bar is genuinely useful and nobody opens it twice a day. Decide which you want before starting.

STACK
- A PWA that works entirely offline, installable, with widgets where the platform allows — or native for the health integration and the watch
- Local storage first: IndexedDB or SQLite on device
- An optional sync server: Node 20+ with Fastify and SQLite
- Caddy in front

THE DATA MODEL
- drinks: id, name, icon, volume_ml, hydration_factor, colour, position, is_custom — a coffee hydrates less than water, and a beer negatively; the factor is what makes the total meaningful
- entries: id, drink_id, volume_ml, hydrated_ml, local_date, at, source — one row per drink, and the local date is the key
- goals: id, base_ml, weight_kg, activity_adjust_ml, climate_adjust_ml, effective_from — appended, so changing your goal does not restate last month
- reminders: id, kind, from_local_time, to_local_time, interval_minutes, is_smart, is_active
- Totals computed from entries at query time. Never stored

THE INTERACTION
- One screen. The day's progress, and a row of drinks to tap. One tap logs the default size, a long press picks another
- Under three seconds from picking up the phone to logged, or it will not happen eight times a day
- Undo the last entry, prominently, because a mis-tap is common and a wrong total is discouraging
- Backfill an earlier time, because people log in batches

THE GOAL
- A sensible default from body weight, with adjustments for activity and heat. State the formula and its source rather than inventing a number
- Adjustable by hand, always. This is not medical advice and the interface should not pretend otherwise
- A note in the README: extreme over-drinking is genuinely dangerous, so cap the goal at something sane and do not gamify exceeding it. A tracker that rewards more and more water is a tracker with a real hazard in it

THE PART THAT IS DESIGN, NOT CODE
- Something has to happen when you log a drink. A character filling, a level rising, a small animation — that is why the application gets opened rather than ignored
- Build it as an SVG or canvas scene driven by one number between zero and one, so the illustration can be replaced without touching the logic
- Respect `prefers-reduced-motion` completely, with a still representation that still reads at a glance
- Commission the artwork or use a permissively licensed set, and credit it. Do not copy the original's character — that is somebody's work

REMINDERS
- Between chosen hours, at an interval, on the days you want
- Smart mode: skip a reminder if you have already drunk recently, and space them by how far behind the goal you are. A fixed reminder every hour is one people switch off in a week
- Never after the goal is reached
- At most one an hour, and silent overnight

HEALTH INTEGRATION
- Write intake to the platform's health store so it appears alongside everything else, and read back anything logged by another application so the total is not wrong
- Read workouts to adjust the day's goal upward
- Read-only where it can be, on device, never uploaded. Health data is the most sensitive thing on a phone and there is no reason for it to leave

WIDGETS AND THE WATCH
- A home screen widget that logs a glass without opening the application. This is where a tracker like this is actually used, and without it the tool is much weaker
- A watch complication with the day's progress and a tap to log
- If your platform will not allow these, say so in the README, because it is most of what the paid application delivers

WHAT IT SHOWS
- Today, with the goal and what remains
- A week and a month as a simple chart, and the average
- Streaks of days at goal, gently — never a notification that scolds a missed day
- Everything computed at query time

SYNC AND ESCAPE
- Optional, end-to-end encrypted, and entirely unnecessary for the application to work
- Export every entry as CSV in two taps
- No analytics in the application. Not one event

OPERATIONS
- If there is a server: .env with DATABASE_PATH, BASE_URL, SESSION_SECRET; migrations on boot; nightly backup of ciphertext

WHAT MATTERS MOST
The one-tap log and the widget. Get logging under three seconds and put it on the home screen, then decide honestly whether you can make it pleasant to look at — because that, not the arithmetic, is what the few euros were buying.

Give me the app, the sync server if you want one, the animation driven by a single progress value, and a README with build steps, the artwork licensing, and the safety note about over-drinking.