still working on this · jun 2026 - present
Marathon Prep Bot
I wanted one place that could remember my training, put each workout in context, and help me tell whether I was actually improving. So I built it around my own workout history.
My runs now update one place where I can track progress, spot patterns, and plan what comes next.
Training apps can tell me how far I ran, but I wanted something that remembered the bigger picture. I wanted it to know the difference between a hard run and a hot run, remember the last few months, and explain what it was seeing. So I built it around my own training archive.
HealthFit exports workout files from my Apple Watch and Polar H10. A TypeScript pipeline deduplicates and normalizes them, then derives heart-rate zones, TRIMP training load, aerobic decoupling, weekly volume, and long-run progression. That context feeds a CLI coach instead of disappearing inside a fitness app.
The public fitness page is another output of the same system. It shows the progress without turning the page into a technical report. Its sync step uses a strict field whitelist: useful activity summaries are published, while GPS coordinates, source filenames, health notes, and workout descriptions stay local.
My favorite part is that the project grows along with the training. Every run gives it a little more context, and every improvement to the project helps me understand the next run a little better.