Project Log
A running record of what changed in the tracker, the placement test, and the plan itself. Newest first.
Set up a permanent link for project updates. Build changes and plan revisions get posted here, newest first, so there's one place to check rather than digging through chat history.
The link never changes — it updates in place.
The tracker went from device-local storage to a proper backend. Her tablet, your phone and any laptop now see the same records.
It stays offline-first, which matters at the mosque: entries save locally with no signal and push automatically when the connection returns. The pill in the top right shows Saved, Saving or Offline.
Sync merges rather than overwrites, so two devices used offline at the same time both keep their entries. Access is a single family passcode with a 90-day session, so she isn't typing it daily.
Verified against a live database: unauthenticated requests are refused, forged session cookies are rejected, resending data creates no duplicates, and an entry written while offline reached the server on reconnect.
CACHE in sw.js. Skip it and already-installed devices keep serving the old cached version, so changes appear to do nothing. Not needed for API changes.
Packaged the app for the free tier: installable to her tablet's home screen with its own icon, opens full-screen, and works with no internet.
Added strict security headers so nothing third-party can ever run in an app she uses daily, and a noindex tag so it won't appear in search results.
The first version had only five questions above Year 6 — so for a child already working ahead, it couldn't tell Year 6 from Year 8. The result would have hinged on a single question.
Now 50 questions spread across Years 4 to 8, and the placement logic was checked against five simulated ability levels to confirm it actually discriminates. Every numeric answer in the key was machine-verified.
The report names every topic she missed — that list becomes the teaching plan.
Seeing the Manchester Central Mosque Academy book corrected several assumptions. The most important: her madrasah says dhor, not manzil. The app and the plan documents now use her terminology throughout.
Also added, to match the book exactly:
Entry is now one daily record rather than three separate logs — under a minute each evening.
Hifz logging with a 30-juz health map, maths logging against a GCSE countdown, a reading shelf, and a Chase-style quiz engine with four banks — prophets and Islamic history, world history, general knowledge, and generated mental maths.
The juz map is the part that earns its keep: it colours all 30 juz by recent mistakes, and shows in pale purple any juz she's memorised but never logged revision for. That decay is invisible in a paper book.
Weekly rhythm, hifz system, the 23-month GCSE maths roadmap to June 2028, and reading and enrichment.
The capacity analysis produced two findings worth repeating. Weekday evenings are unavailable — she wakes at 05:30 and must sleep by 20:00, so anything after 19:30 comes out of her sleep, and tired sabaq costs more than the maths gains.
And the mosque car journeys are what makes the plan fit: 50 minutes each way, five days a week, currently empty. Without them the maths hours don't add up to a GCSE.