testing
Topics · 4 entries
Lab
An automation stack built by AI agents, under tests
Roughly thirty scheduled jobs nobody watches, written mostly by AI agents — which only works because every job has a test that runs it, and the tests get broken on purpose to check they can still fail.
Projects
An AI-run home lab
A home network and automation stack where the agents do the building and the tests decide whether it counts.
Writing
Directing AI to build, and how you know it worked
An AI will tell you the work is finished in the same confident voice whether it is or not, so the job of the person directing it is to decide, before any work starts, what "finished" will be checked against.
A test that has never failed proves nothing
A green check only means something if you have seen it go red. The cheapest way to find out is to break the thing it guards on purpose and watch.