# Two MPH

> A workday planner for an under-desk treadmill (a slow walking belt under a standing desk) at https://two-mph.skillsafe.ai/. Paste one day from your calendar or drop an .ics file; the page sorts each block into typing, talking, listening, break or other and drafts a program - walk, stroll, sit or off the desk, with a speed and the minutes to walk - held under a ramp that depends on how many weeks you have walked at your desk. A treadmill-style console shows the program as an LED dot-matrix profile with walking minutes, distance, estimated steps and the longest sitting stretch, and a 4-40 sitting-risk score is shown before and after. A metered coach run (model gpt-terra) reviews the day block by block. Built on the SkillSafe skill @huifer/occupational-health-analyzer.

## What is free (in the browser, no sign-in)

- Calendar parsing: "09:00-09:30 Standup", "9am 30m Inbox", "1-2pm Client call (camera on)", times without am/pm before 7 read as afternoon; overlaps trimmed; up to 24 blocks in 14 hours. An .ics export is read for its busiest date.
- Block kinds from the title: typing (write, code, email, inbox, report, admin...), talking (present, pitch, interview, client, workshop, coaching...), listening (standup, sync, 1:1, all-hands, webinar, review, reading...), break (lunch, gym, errands) and other; "camera on" is noted.
- Ramp levels by weeks already walking at the desk: new - 60 walking minutes a day, 20 at a stretch, top 1.5 mph; about a week - 90 / 30 / 2.0; two to three weeks - 120 / 45 / 2.5; a month or more - 180 / 60 / 3.0. Never above the treadmill's own top speed. These are conservative planning defaults, not a published standard.
- Three programs (Gentle, Steady, Mileage) decide which kinds walk; the day's walking minutes are spread across the walkable blocks; you can change any block's call and the limits still apply.
- Console numbers: walking minutes, distance, estimated steps (cadence about 58 + 20 x mph steps a minute - an estimate), longest sitting stretch.
- Sitting-risk score from the source skill: workday sitting hours, how often you get up, weekly exercise and discomfort, each scored 1-10, for 4-40 (low 4-13, medium 14-26, high 27-40), shown now and with the plan.
- Stop words: if the notes mention a current injury, sprain, surgery, swelling, dizziness, chest pain, a clinician saying rest and similar (or discomfort is severe), the page plans the day seated.
- Exports: Markdown, CSV of the program, an .ics file of the walking blocks on a date you pick (default: today, or the next workday once today's first block has started), JSON of everything. Three sample days replay free.

## The metered lane

`task: "program"` - one coach review. Input fields (all strings; objects JSON-encoded): `blocks`, `profile`, `notes`, `page_program`, `page_flags`, optional `retry_note`. Reply: `{task, verdict, headline, summary, blocks: [{id, mode, speed, minutes, why}], moves: [{what, why}], ramp: [{week, daily_minutes, longest_bout, top_speed_mph}], setup, stop_if, seated_breaks}`.

Verdicts, decided in order: `sit_first` (a current health matter in the notes that walking could worsen or that needs an all-clear, or severe discomfort - every block seated, no ramp, seated breaks instead), `ease_in` (new or about a week of desk walking), `walk_ready` (two weeks or more). The page re-checks every reply: one call per block, speeds under this week's top, the daily total and the longest unbroken stretch, the verdict rule, the ramp starting at this week's limits, no internal names in prose.

## Limits and honesty

- Not medical advice. The app screens and plans; it never diagnoses. Anyone with pain, an injury or a health condition should ask someone who knows their health before walking at a desk.
- Steps, distance and the risk score are estimates; the console's numbers come from the program, not from a device.
- Not affiliated with any treadmill maker.

## Pages

- App: https://two-mph.skillsafe.ai/
- API tutorial (eight languages, worked examples for every verdict): https://two-mph.skillsafe.ai/api.html
- Source skill: https://skillsafe.ai/skill/@huifer/occupational-health-analyzer