Stack Diary — Requirements

Goals

• Give supplement users a simple daily logging habit that takes under 60 seconds.
• Surface which specific supplements correlate with better energy, sleep, mood, and focus in the user's own data.
• Provide adherence accountability through streaks and progress visualisation.
• Be runnable in any browser with no server, build step, or account required (prototype phase).

Primary user

Alex, 34, product manager and biohacking hobbyist. Spends ~$150/month on supplements after falling down the Huberman Lab/Bryan Johnson rabbit hole. Takes 8–12 supplements at various times of day. Sometimes forgets evening supplements. Has no systematic way to tell whether the stack is working or which specific items are moving the needle — just a vague sense that "feeling better lately." Wants evidence, not guesswork.

Functional requirements

• FR1: Display a supplement stack as a card grid showing supplement name, dose, timing, category, 30-day adherence %, adherence progress bar, and current streak.
• FR2: Compute per-supplement adherence percentage from log data (days taken / total logged days).
• FR3: Compute per-supplement day streak (consecutive days from most recent where supplement appears in taken list).
• FR4: Display the last 14 days of logs as a table with adherence count, per-outcome dot visualisation (1–10 scale), and composite wellness score.
• FR5: Render a multi-line Chart.js chart showing 30-day trends for energy, sleep quality, mood, and focus.
• FR6: Compute supplement → outcome correlations by comparing average outcome scores on days a supplement was taken vs. not taken.
• FR7: Render the top correlation findings as human-readable insight cards (e.g., "Magnesium Glycinate → Sleep Quality +2.8 pts").
• FR8: Only surface correlations where both sample groups have at least 2–3 data points and the difference is ≥0.5 points.
• FR9: Display a sticky header with three live-computed stats: current streak (≥70% adherence per day), 30-day adherence %, and average composite wellness score.
• FR10: Implement tab navigation between My Stack, Daily Log, and Insights views.
• FR11: Load all data from an embedded inline JSON block (no fetch required, works from file://).
• FR12: Apply protocol name from data to the Stack tab heading dynamically.

User stories

• As a health optimizer, I want to see each supplement's 30-day adherence at a glance so I know which items I'm consistently forgetting.
• As a protocol follower, I want to see my current streak so I stay motivated to maintain daily consistency.
• As a self-experimenter, I want to see which supplements correlate with better sleep so I can decide whether to drop underperforming ones.
• As a busy professional, I want the log table to show the last 2 weeks at a glance without scrolling through a calendar.
• As a data-driven person, I want a trend chart of my four wellness metrics so I can see whether my protocol is improving outcomes over time.
• As a new user, I want the UI to immediately make sense without any tutorial — card layout and tabs are self-explanatory.
• As a mobile user, I want the layout to reflow cleanly on a phone screen so I can check my stats on the go.

Non-functional requirements

• Performance: All rendering must complete in under 200ms on a modern browser with 30 days of log data and 10 supplements.
• Accessibility: Color is not the only differentiator — scores also show numeric values alongside dot visualisations.
• Privacy: All data is local (embedded in the HTML or a JSON file). No analytics, no tracking, no network calls beyond the Chart.js CDN load.
• Offline: Works with no internet after the Chart.js CDN script is cached.
• No auth: The prototype requires no login, account, or API key.
• No build step: Pure HTML + CSS + vanilla JS; open index.html directly.

Out of scope (for the prototype)

• User authentication or cloud sync
• Adding or editing supplements within the UI
• Daily check-in UI (data is pre-loaded from JSON)
• Push/email reminders
• Wearable integrations (Oura, Whoop, Apple Health)
• Protocol template library
• Export or sharing features

Open questions

• Should the correlation engine require a minimum of 5 "without" days before surfacing a finding, to avoid false positives from low-variation protocols?
• What is the right minimum diff threshold — 0.5 points or 1.0? (Lower captures subtle effects but risks noise.)
• Should outcome scores be self-reported sliders (1–10) or derived from wearable data? Both paths have monetisation implications.
• Is "wellness score" (mean of four outcomes) the right composite, or should users weight metrics they care most about?