Vol. I · No. 579Rochester, New York, Thursday, September 24, 2026Productivity · Page B26
All workSection B · Page B26

ProductivityHabitTrack

A Habit Tracker That Understands Every Third Day

A small habit and weekly time-log tracker with custom recurrence, automatic weekly resets and nothing stored on a server.

HabitTrack, a small habit and time tracker
HabitTrack, a small habit and time tracker.Screenshot: habitrack-brown.vercel.app

Habit apps tend to assume every habit is daily. Some of mine aren't. In HabitTrack each habit has its own interval, and a completion covers that whole window, so a gym day every other day doesn't show up as a broken streak.

The second tab logs hours against weekly targets. Every Monday the week resets on its own and moves into history. Data stays in localStorage, with JSON export and import for backups, and two of the commits were timezone fixes so days roll over at local midnight. The whole app is one 734-line React component, and it's deployed on Vercel.

More from The Work

See all 25

Information theory

A Wordle Clone With an Entropy-Powered Cyborg Sidekick

A full-stack Wordle clone with a Cyborg Mode that suggests the best next guess using Shannon entropy.

A complete Wordle clone in React with a twist: flip on Cyborg Mode and a Flask solver suggests the guess that splits the remaining answers the most, measured by Shannon entropy.

Continued on Page B21

Design

A Writing App That Feels Like a Printed Book

A distraction-free personal editor with cream paper, drop caps, running headers and page numbers.

Folio is a personal writing app for stories, lessons and thoughts that looks like writing inside a real book: a two-page spread, IM Fell English type, drop caps and page numbers. The toolbar hides until you select text.

Continued on Page B19

Fintech

A Local Llama Audits the Company Spending

A team-built spend auditor where a local Llama 3 model checks employee transactions against company policy and flags risks by level.

SU$PEND lets an auditor pick an employee, pull their transactions and run an AI audit against the company spending policy. A local Llama 3 model does the checking, so financial data stays in-house. I built the audit pipeline and the backend.

Continued on Page B16