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.
Decisions that are pending and would block or shape work — each naming what would resolve it. Nothing else lives here: known answer + small work goes to the punch list, known direction goes to Future Considerations. At phase open you read your area's questions, not all of them.
2 high priority · 3 topics · areas match the feature registry
Thinking: This is the riskiest assumption in strategy/Scope & Constraints.md. If the camera cannot separate "bright indirect" from "medium" across different phone models, the room survey becomes a questionnaire and the product's premise weakens badly. The sweep is queued.
Resolves when: the sensor accuracy sweep reports agreement against a real lux meter across at least four handsets.
Thinking: Overrides would rescue us where the camera is wrong, but they also let a user re-create the generic advice that kills plants. Leaning toward allowing it with the reasoning still visible, so the override is an informed one.
Resolves when: decided in the Watering reminders walkthrough, or deferred with a trigger.
Thinking: Care history is queued partly to build streaks, but a broken streak is exactly the moment our "copy never scolds" principle gets tested. A plant that died is not a lost badge. Worth designing the failure state before the success state.
Resolves when: the Care history phase states its position, and copy is reviewed against the Vision principle.
Thinking: Trivial while there is one user per home, and it becomes a real question the moment shared households arrive. Recording it now so the schema choice is deliberate rather than accidental.
Resolves when: shared households are scoped, or the notification schema is next touched.
Thinking: Every room survey becomes invalid at once. Silently keeping old schedules is wrong; invalidating everything without warning is worse. Probably a prompt, but the trigger for detecting a move is unclear.
Resolves when: a phase owns re-survey prompting, or it is promoted to a Future Consideration with a trigger.
Source: planning/Open Questions & Assumptions Log.md — read your area's questions at phase open