n1lab — changelog

14 may 2026

0.3.4

— "field notebook"

beta

Trial close-out, redrawn.

The 24-hour trial outcome screen is the one place the app does the most reasoning, and it was the loudest screen in the product. This pass quiets it down.

changed

Trial close-out now uses the three-window model (0–2h, 2–6h, 6–24h) end to end. The 5-day model is gone — every chart, every prop, every label.

changed

DeltaStrip charts draw the 14-day baseline as a shaded band, not a line. A reading inside the band is unremarkable; outside is the narrowest definition of "something happened."

added

Confounders recorded at trial start now surface in the close-out — including a free-text note and the source (manual or suggested).

fixed

Confidence line no longer rounds the agreement count. "3 of 4" stays "3 of 4."

Method note · § 04

The earlier 5-day window was a compromise from the original PRD. Three trials and a clinician's eyebrow later, we cut it. The intra-day windows match how reintroduction reactions actually present.

21 apr 2026

0.3.3

— "compendium"

beta

Cross-trial rollup, two views.

A food can be trialed more than once. The app now treats that as the normal case.

added

FoodSummaryCard — a single card that rolls up every trial for a given food, with the trial outcomes laid out horizontally and a confidence line beneath.

added

Today now shows an inline prior-trials strip on the active trial card when it's a repeat. Surfaces the history without leaving the screen.

changed

Vocabulary review. "Pattern detected" replaced everywhere with "possible association." See vocabulary.

30 mar 2026

0.3.2

— "weekly review"

beta

The weekly pass.

Sunday-evening review of the previous seven days. The app surfaces what it noticed; you decide which observations to keep.

added

Weekly review screen with callout cards for each surfaced observation. Each callout shows the observation, the evidence count, and a confidence line — never a conclusion.

added

Inline "save as observation" action on each callout. Saved observations carry forward into next week's baseline computation.

changed

Notes from the day's free-text field now thread into the weekly review surface, with timestamps preserved.

removed

Streak indicator on the weekly review header. Was a holdover from a much earlier sketch; should not have shipped at all.

04 mar 2026

0.3.1

— "burden grid"

beta

Symptom sliders, anchored to you.

Each symptom domain slider now references your recent range rather than a universal 0–10. A score of 6 means the same thing it meant last Tuesday.

changed

SymptomBurdenRow redraws against personal baseline. Anchor ticks at the 14-day mean and at ±1σ.

fixed

Null vs zero — a skipped domain is now stored as null, not 0. Daily mean computation respects nulls.

added

Tap-and-hold on a symptom row reveals the value the score replaces (so you know what you were comparing against).

06 feb 2026

0.3.0

— "instrument"

beta

A pass on the visual system.

Everything is hairlines now. The app feels less like a consumer product, more like a small piece of laboratory software. That was the intent.

changed

All cards moved from shadow-based elevation to hairline borders + layered surfaces. Dark mode is the default; light mode is secondary.

changed

Icon set standardized to Lucide at 1.5 stroke, monochrome. Custom glyphs removed.

removed

Every gradient on a surface. Every confetti animation. Every emoji.

Design note

This release marks the point where n1lab stopped trying to look friendly. The product was always supposed to be quiet; the visual system caught up.

12 nov 2025

0.2.0

— "first trials"

alpha

The 24h food trial, end to end.

First end-to-end flow for reintroduction trials, from setup to close-out. Initial three-window engine.

added

ChallengeSetup, ChallengeOutcome screens. Trials persist across sessions; intra-day windows surface their own prompts.

added

Trial outcome vocabulary — tolerated, reaction observed, inconclusive, aborted. No fifth option.

28 aug 2025

0.1.0

— "the leaf"

alpha

First leaf in the notebook.

Minimal AM and PM check-ins, the eight symptom domains, weight, and a single trend chart. The shape of everything that came after.

added

AM check-in, PM check-in, symptom burden across eight domains.

added

Weight logging plotted against a chosen baseline.

added

A single line chart for daily burden. Initial 14-day baseline computation.