Example project — Frond is not real.
Authored demo content for a fictional plant-care app, rendered by the same parsers and held to the same invariants as a real project's dashboard. This is the system in use, three months in.
Roadmap
The boards open now, the queue of planned work of any mode, and where the project stands.
Open boards
Where we are
Onboarding, plant profiles and the room light survey have shipped. A new user can add a plant, tell us where it lives, and see a care profile that reflects the room rather than a generic species average.
Watering reminders is the open phase and the first thing that makes the app worth opening twice. The light survey is what makes its schedules defensible, which is why it shipped first.
After reminders land, the question becomes retention: care history is what turns a reminder into a habit, and it is queued next.
History lives elsewhere: decisions.md (why) and the phase records in archive/phases/ (what shipped).
The queue
A per-plant timeline of waterings, so a reminder becomes a streak worth keeping.
Close the gap between phone-camera light readings and a real lux meter.
Split the growing implementation folder and retire two merged references.
Key considerations
Things to keep in mind across phases. Not tasks — lenses.
every phase that adds a push has to argue why it is not the one that gets us muted.
we license it, we do not curate it, and we never present it as certainty.
seasonal logic is hardcoded to one hemisphere and will embarrass us the first time someone in Melbourne uses it.
On the horizon
- Shared households, where two people water the same plant.
- A plant-identification camera flow, once the care model is trusted enough to be worth pointing at a stranger's monstera.
Full doc: ROADMAP.md