DEV Community

Philip D'Souza
Philip D'Souza

Posted on Originally published at healthexport.dev

Setup time is the pillar nobody puts on an Apple Health connector compare page

Every Apple Health connector compare page I have read ships the same three rows: a metrics count, a tools count and a privacy line. Almost none ships a time row, and time is the pillar that decides whether your agent answers tonight or next weekend.

Here are the numbers, taken from each project's own documentation rather than a marketing claim.

Route Their own number What the clock covers
Hosted MCP connector under a minute paste a URL, complete OAuth, no local install
MetricBridge (export plus local server) about two minutes export to a folder, one install line, restart the client
health4.ai about 15 minutes create the destination, connect it, wait for a background sync iOS schedules
Open Health Sync about 20 minutes Docker Compose stack, create a user, import into PostgreSQL, mint a token, add the server

The spread is 10x and no compare page prints the row.

Why the number never makes the page

Setup time is not a spec-sheet value. It only exists after someone runs the install on a real machine, so a page built from a README cannot copy it in. It is also hard to self-report without sounding slow: "about twenty minutes" is honest and reads like a warning. And minutes are not a screenshot. A tool count photographs well. A stopwatch does not.

The part the minutes hide

Setup is a one-off. Maintenance is the bill. Read the twenty-minute route closely and you find chores that never show up in a time-to-first-answer number:

  • a read-only token that expires every 30 days
  • a container stack that has to stay running
  • a database you now back up
  • a background sync iOS schedules, not you

The local-file route has none of that. There is no token to renew, because the grant is the folder path. There is no service to keep alive, because the server is a process the client starts on demand.

Benchmark any route yourself

Four rules that make the numbers comparable:

  1. Same machine, same client. A cold laptop with no Node is a different test.
  2. Start from nothing. No export sitting around, no Docker installed.
  3. Stop at the first real answer. Not the first green check.
  4. Count the moving parts, not just the minutes.

Do that once per route and you have a column no compare page gave you. Point it at whichever project you test, ours included. If our two-minute number does not survive your stopwatch, that is worth knowing.

The full breakdown, with the fair-benchmark rules, is on our blog: https://dub.sh/X1klFsp

Top comments (0)