LSAT Tracker

A single-file study tracker for LSAT prep: daily schedule and minutes, a wrong-answer journal, section and practice-test scores, and miss-pattern analysis — split by Logical Reasoning and Reading Comprehension.

Built to replace a spreadsheet, so everything is one HTML file with no build step.

Tabs

Tab	What it does
Today	The day's scheduled task, a minutes box, week progress against the weekly goal, streak and totals
Day log	Every past week, grouped Monday–Sunday, with per-day tasks and minutes; backfill any missed day
Wrong answers	Log a missed question with type, level, time, PT/section/question, and both halves of the reasoning
Scores	Sections, drills, and full practice tests, with an accuracy line for sections
Patterns	Misses by question type and difficulty, average time on a miss, and the blind-review gap

An LR / RC toggle carries across Wrong answers, Scores, and Patterns. Full practice tests sit outside that split, since a PT score isn't one section type.

Running it

index.html works three ways, and detects which one it's in automatically, in this order:

1. Inside a Claude artifact. If window.claude.use("db") resolves, the page uses the Claude artifact host's db capability. Data lives there, synced across every view of that artifact. This is the private, edit-in-conversation copy.
2. Vercel + Neon. If this deployment has a /api backend (see below), the page talks to it over fetch. Data lives in a real Postgres database — synced across every device and browser that opens the URL. This is the one meant for actual day-to-day use.
3. Standalone, no backend — GitHub Pages with no /api, a plain web server, or straight off disk. Falls back to a localDb() shim backed by the browser's localStorage. Data then lives per-browser, not synced anywhere — clearing site data or switching browsers starts you over.

All three implement the exact same doc/collection/onSnapshot/get/set/delete/add shape, so every render/write function in the page runs completely unmodified regardless of which one is active.

"Back up data" exports every collection as one JSON file — through the artifact host's downloads capability where available, or a plain <a download> blob otherwise. "Restore backup" reads a file of that same shape back in: tasks and journal/score entries with a matching date or id are overwritten, everything else is added. Backup/restore is how you move data between modes — e.g. export from the Claude artifact, then restore into the Vercel copy.

Deploying to Vercel + Neon

1. Create a Neon database. New project at neon.tech → open its SQL editor → paste in and run schema.sql from this repo. Then copy the connection string (Dashboard → Connection Details — use the pooled one).
2. Import this repo into Vercel. vercel.com/new → import LSAT-Tracker from GitHub. Framework Preset: "Other". No build command needed.
3. Set the environment variable. In the new Vercel project → Settings → Environment Variables → add DATABASE_URL = the Neon connection string from step 1. Redeploy if the project already built before this was set.
4. Open the deployed URL. The page's own boot sequence detects /api/config and switches to the Neon-backed store automatically — nothing else to configure.

The API is five plain Vercel serverless functions under api/ (no framework), using @neondatabase/serverless — see package.json for the one dependency.

Data model

Four collections:

• meta/config — weekly goal, start date, and the baseline totals carried over from the original spreadsheet (minutes, days completed, best streak, and per-week minutes for weeks that predate the tracker)
• tasks/<YYYY-MM-DD> — one document per scheduled day: title, cat, plannedMin, skip, done, actualMin
• journal/<id> — date, mode (LR/RC), source, qtype, level, timeSec, test, section, q, whyWrong, whyRight
• scores/<id> — date, kind (LR/RC/Drill/PT), mode, raw, total, timeSec, br, note; full PTs carry scaled instead of raw/total

Dates are YYYY-MM-DD strings parsed as local time. Weeks run Monday to Sunday.

Structure

Everything is in index.html — styles in a <style> block using CSS custom properties with light/dark variants, and an IIFE holding state, date helpers, render functions per tab, and the write functions. No dependencies beyond two Google Fonts.