April 14, 2026

Overview

Attendance Tracker runs events for a Singapore financial advisory group. Members register for events and check in with a QR code. Managers approve leave and excusals. Admins see expected, missed and excused counts for each event and agency. They also send email reports, calendar invites and reminders. It ships as a web app and a Capacitor mobile app.

My role

Co-developer for four weeks, from March to April 2026. I joined a client product that Lovable built most of. The client started it in Lovable in July 2025, and another developer worked on it before me.

• 91 commits on 13 days
• 91 of the 101 commits people made from my first day on
• Lovable made 97% of the repo's 7,055 commits
• 23 of my commits changed what users see. 47 fixed the data logic behind the screens.

Registration, gamification and the mobile QR app came before me. My work was the attendance logic: who counts as expected, absent, excused and reminded.

What I built

• Expected, missed and excused counts: fixed across analytics, engagement, missed events and leave views.
• Absentee computation for follow-up events, across every audience mode and chained source events.
• Manager leave approval aligned with the admin flow: a table view, per-event excusals, and scope and eligibility checks.
• Excusal fix: one member's approved leave no longer excuses everyone in a batch check.
• Amend Leave: agency managers can amend agency-scoped excusals on approved leave.
• Calendar invites and reminders that reach the right people, including custom groups, pinned users and absentees.
• Email event reports: an attendee details table, a Resend button, a collapsed history, and correct expected and guest counts.
• QR sign-in popup for logged-out members who scan a check-in code. The redirect survives sign-in.
• Smooth expand and collapse on analytics event rows and the per-agency breakdown.
• Create-event form: location suggestions from past events and meeting link suggestions.
• User forms: calendar date pickers in the create and edit user modals.

Lovable built the analytics pages. I fixed the counting behind their columns.

How I built it

I built my part with Claude Code. 71 of my 91 commits credit Claude as co-author.

• In April 2026, I committed the agent setup: 3 subagents, 10 skills and 2 commands.
• A tag on each task sets its checks, such as a research step or a Playwright check.
• A /batch command works the task queue on its own.
• The setup includes reviewer and QA subagents.
• Agents do not apply database changes. They go into a handoff file for the client and Lovable to apply.

Tech stack

• Web: React 18, TypeScript, Vite, Tailwind CSS, shadcn/ui, TanStack Query, React Hook Form, Zod
• Mobile: Capacitor 7 for iOS and Android, with a barcode scanner
• Backend: Supabase (Auth, Postgres, 35 Edge Functions), Resend for email
• Maps: Google Maps, Mapbox
• Tests: Vitest, Playwright
• Agents: Claude Code, alongside the client's Lovable