distance-nerd-stuff

seriously, who cares?

I do, apparently. This is a little personal corner of the internet for poking at my own endurance sports data: a Strava dashboard for activities pulling using the Strava API activities (2024+), and a Running Log: my college running log that predated Strava entirely (2003-2007).

Live:

• 🚪 Landing page — https://ducktapegirl.github.io/distance-nerd-stuff/
• 🏃 Running Log — https://ducktapegirl.github.io/distance-nerd-stuff/college.html
• 🚴 Strava dashboard — https://ducktapegirl.github.io/distance-nerd-stuff/strava.html
• 🖼 E-paper displays — the fridge panels, one card an hour
  • Today's card, big gray panel — https://ducktapegirl.github.io/distance-nerd-stuff/epaper.html
  • Today's card, small four-color panel — https://ducktapegirl.github.io/distance-nerd-stuff/epaper_xiao.html
  • Every haiku so far — https://ducktapegirl.github.io/distance-nerd-stuff/haiku.html

What's actually here

• Strava dashboard — charts and stats built from my Strava activity history: pace trends, segment performance, mountain bike speed, that kind of thing. Refreshed automatically a few times a week.
• Running Log — my running history going back well before Strava existed, parsed out of old hand-kept HTML logs into one browsable, searchable page.
• An e-paper feed — the same data cut into 63 single-fact "cards" for a little 800×480 gray-scale panel stuck to the fridge. Sixteen of them rotate, one an hour. No color, no JavaScript, nothing smaller than 26 px, because at 235 PPI the whole screen is about the size of a credit card. The concepts for all of these were wholly generated by Claude (Opus 5 and Fable 5.1). I chose from about 100 candidates.
• …and a second, much smaller panel — the same sixteen-card rotation redrawn for a 2.9″ 296×128 display that has four colors instead of four grays: black, white, red and yellow, no dithering and nothing under 14 px. Same data, same hour, a tenth of the room, so most cards had to be drawn again rather than scaled down.

All of it is static pages, rebuilt from data + a few Python scripts, and published with GitHub Pages.

Built by a team of robots (sort of)

The Strava dashboard is my attempt at a subagent + orchestrator framework. It's built and maintained by a small crew of Claude subagents, each with one job: one decides what's interesting in the data, one designs how a new chart should look, one writes the actual code, one checks the result before it ships. It's equal parts "I wanted these specific charts" and "I wanted to see how far an agentic build pipeline could go." Curious how it works under the hood? See strava-data/AGENTS.md.

Filing an issue

Want something fixed or added? Open an issue. There are three forms, each with different tagging and metadata to help me process input:

• Bug report — something's broken, wrong, or looks off. This form auto-tags itself bug.
• New view / chart idea — propose a question you want a dashboard to answer, or a fact you'd like an e-paper card to carry. This form auto-tags itself enhancement.
• General enhancement — anything else you'd like improved that isn't a new chart or view (navigation, layout, exports, workflow, etc.). This form auto-tags itself enhancement.

All three start by asking which page or panel you mean: either dashboard, the landing page, or one of the two e-paper panels.

Filing an issue with the bug or enhancement tag doesn't trigger anything by itself. Only I can approve a triage action. Once I approve an issue, then it can get picked up and routed automatically: bug reports go straight to a fix, and enhancement requests go through the full design pipeline if rough, or get built more directly if specific (based on your own "how formed is this idea?" answer on the form). You'll see progress as label changes (agent:in-progress → agent:done) and eventually a pull request that closes the issue.

Can I use this?

Not as-is. The data here is mine — the Strava dashboard is built from CSVs pulled through my personal Strava API credentials, so you can't just clone this and get a working dashboard with your own data. But the build scripts themselves are plain Python and free to read or borrow from. How it's all built and run lives in CLAUDE.md.

What's next

A handful of ideas are written up but not built yet — a real heat-stress index for the heat-vs-pace charts, clickable links out to Strava, records that update themselves when a bigger hike comes along, and an orphaned summer-2003 log the parser has never read. They're collected in Project Docs/Plans/README.md.