Sunnsteel

Supabase is the identity provider, but it is not the authority the API trusts blindly. Supabase issues a JWT; a custom Passport strategy on the NestJS side verifies that token on every request, locally against the project’s published signing key so no request waits on Supabase, and upserts the caller into the local Postgres users table, so application data has a real foreign key to point at rather than a floating external id.

The frontend keeps an HttpOnly cookie as a session marker. Next.js middleware reads it to decide whether to render a protected route or bounce to sign-in. The cookie is deliberately not the credential — see the engineering note below.

The wizard runs basic info, then training days, then a per-day exercise builder, then a review step before anything is written. Each exercise is configured with a set count, either a fixed rep target or a rep range, rest seconds, and free-text notes.

Exercises come from a catalogue tagged with primary and secondary movers plus equipment metadata, which is what lets the app attribute volume to muscle groups rather than just counting sets. Finished routines can be favourited, duplicated, edited or marked complete.

A routine is meant to outlive a training cycle. Saving a version captures it as it is, to compare with the current setup day by day and restore later. Dated training blocks plan named, non-overlapping phases from the routine or one of its versions, and while a block is in force the schedule, the dashboard and new workouts follow it. A deload plans up to two weeks of lighter loads, previewed exercise by exercise, without touching the routine or its blocks. Adjacent exercises link into supersets and circuits that keep their own sets and progress.

Routines also travel between members. Start from one of three curated templates; share a routine with followers, with everyone, or through a revocable private link, which carries the programme and never your workouts or notes; and clone anyone’s into a copy of your own that remembers whose it was. Shared routines can be browsed by goal, level, days a week, muscles trained and the equipment you actually have.

Every set row carries its own save state — idle, pending, saving, saved or error — so a failed write is visible on the row that failed instead of behind a single global spinner. That matters in a gym on bad reception, which is the normal case rather than the edge case.

Sessions persist and can be resumed, and one abandoned mid-workout is offered back for recovery rather than silently ended; they track their own duration and accept notes. History is filterable, and a past session opens into a planned-versus-actual view so you can see where you deviated from the programme.

The session handles the parts of a real gym that a form would not. Warm-up ramps are built from the bar and plates you own and rise with the working weight as it progresses. Sets are marked as warm-up, working, drop or optional, and only working sets move a load forward. Supersets and circuits run in rounds, an exercise can be swapped mid-workout without touching the routine, and the screen stays awake. Rest alerts reach a locked phone through Web Push, naming the lift that is up next. Crossing a weight, rep, volume or estimated one-rep-max record is celebrated the moment it happens, and finishing opens a recap against the last session of the same day.

Three schemes are available. None holds the prescription steady. Double progression adds reps until the top of the range is reached across all sets, then adds weight and drops back to the bottom of the range. Dynamic double progression applies that same logic to each set independently, so a set that stalled does not hold back the sets that did not.

The rule runs server-side on session completion, which keeps a single implementation of the maths and means the next session is already correct whenever and wherever it is opened. Finishing a workout then explains every change it made: which exercise, which rule, the reps performed against the target, and the old and new load.

Pick a trained lift and the app charts its best-set load and estimated one-rep max over thirty days, ninety days, six months, a year or all time. Each range carries in the record that preceded it rather than starting the curve from whatever happens to fall inside the window, and the line is paired with the exact dated load-and-rep changes behind it. None of it uses a charting library: the trend chart is hand-written SVG.

The muscle-group heatmap compares four, eight or twelve weeks of training distribution, counting a completed set fully for the primary mover and half for each secondary one. That is what the catalogue metadata in the routine builder was for: volume gets attributed to muscles instead of being a count of sets. Weeks with no work show an honest zero, and the current partial week is marked as partial rather than quietly compared against completed ones. Beside them sit each exercise’s full performance history, volume trends, session-to-session comparisons, a timeline of records and load changes, private measurable goals, body measurements, a consistency calendar and training-block comparisons.

Plateau watch is the part I would show first. It lists lifts trained at least four sessions since their current best — a threshold you can move between three and eight, saved to your account — where that best is at least three weeks old. Each row gives the best set, its estimated one-rep max, and the closest you have come since. The rule is printed beside the list, and the feature never proposes a cause. It says what happened; the diagnosis is yours.

Training signals hold the same line. Four measures from the last two weeks sit beside the two before: average RPE, sets that fell short of their rep target, lifts whose estimated one-rep max fell twice running, and the number of workouts, each with its window and the threshold that marks it as changed. When they stay marked, Progress offers a short deload for the routine you train most, fills in the dates and lighter loads, and changes nothing until you save it.

The schedule puts each week’s planned routine days next to the sessions that were completed, ended early, or are still in progress, and can switch to a whole month with one mark per day. A rotation that has chosen training weekdays lays its upcoming days out on those weekdays in order; a rotation without them says which day is next instead of inventing a date for it.

The wording is the design. A past planned day with no workout is marked “not logged”, and a planned rest day counts as rest rather than as a gap. An app that tells you that you failed on Tuesday is an app you stop opening in March.

Plans move, so the page lets them: a planned workout can be shifted up to six days without editing the routine — the original day records where it went, and the dashboard follows it — or skipped ahead of time, or skipped after the fact for a day within the last six, or postponed to the next free day in one tap. Every one of those is undoable.

The schedule also reaches you when the app is closed. Pick a time and it reminds you on planned days only, staying quiet on rest days and on days you have already trained; on the last day a streak can still be kept, it says so, with the dates rather than a lecture, in place of that day’s reminder.