Messages waited to be read.
A reply from Spark sat on the board for three minutes before Instinct looked. Nothing was lost. Instinct just checked on a timer. For now, Spark says when it expects to finish, and Instinct looks then.
A doorbell, not a second board.
I decided Spark needed a doorbell to ring, so Instinct wakes up and reads the message on the board. We wrote the safety plan first. By the end of the day it was still only a plan.
I still had to double-check them.
Instinct told me it had sent Spark a request. The request never reached the board. New rule: the assistants coordinate on the board, where I can see their messages, unless I agree otherwise. I still felt I had to check what Spark and Instinct told me, and I can't hand off work until I trust their reports.
Tested on a copy first.
Before touching the live app, Instinct tried every organizer and player step on a test copy of KheloHQ. Instinct found one bug: the setting that lets players enter their own scores was lost when a tournament was created. Spark fixed the bug, and Instinct checked everything again.
A review caught three gaps.
Spark wrote a plan for the release, putting the new changes into the live app. Instinct reviewed it and found three gaps: a missing step to update the app's data, files that did not exist, and tools that only worked on Spark's computer. All were fixed before anything ran.
I said go.
We had a tested backup and a tested way to undo the release. I said go from my phone, with one condition: stop at the first failure, and do not quietly try again.
Spark went quiet.
Spark went silent for 76 minutes in the middle of the release. Instinct checked what it could see: the app was up, and a fresh backup had been taken. Spark was working, just not reporting. We noted the silence as a problem to fix.
Live.
KheloHQ's first version was live. 101 automatic checks on the live app and its data passed. One outdated test failed, but the app itself passed the checks.
