We studied CycleStack before deciding what Supplements should become
A repository study separated useful stack, schedule, history, and supply mechanics from the code and product assumptions Protocol should leave behind.
The research question
Protocol already had access to an earlier supplement-tracking experiment called CycleStack. Reusing it directly would have been fast, but speed was not the main question. We needed to know which parts belonged inside a training-led Health OS and which parts would import the wrong architecture, interaction model, or product promise.
What we would keep
The durable jobs are practical: record a user-authored stack, schedule the choices the user already made, mark planned occurrences as taken, skipped, or unmarked, keep inventory correctable, project when supply may run out, and preserve a refill link.
Those mechanics can fit Protocol's compiler model. A definition plus schedule, intake history, inventory events, date, and timezone can produce today's agenda and a supply projection without depending on the network.
What we would reject
The study found duplicated models and calculators, split intake ledgers, ambiguous timezone behavior, fixed projection horizons, and seeded content that could make the product look recommendation-adjacent. Protocol would replace those patterns with one event ledger, explicit local-date semantics, visible confidence states, offline-safe compilation, and user-controlled sync.
The product boundary stays narrow: no recommendations, no intake instructions, no dosage guidance, no efficacy claims, and no medical or health advice. Protocol would organize choices the user already made; it would not decide what belongs in the stack.
Availability
Supplements is not live in Protocol. This work is an adoption study and owner-reviewed product direction. User research, copy testing, Training evidence, and a pure domain spike must precede any implementation plan.
Next
Test whether inventory, refill, and memory friction are strong enough to justify a Training-adjacent module, and stop if the demand is mainly for recommendations or clinical interpretation.