Why we built it
We run automations on several servers. Fixing them by hand in a browser is slow, and it is easy to touch the wrong one when many share a server. We wanted one place to see, test and change them under strict safety rules.
What it does
It is a toolkit that lets our AI coding assistant list, read, create, debug and update automations across those servers. It drafts one from a plain-language description and checks it before it goes out.
On the shared production server, every change is backed up first and blocked if the backup fails. New automations are created inactive until we activate them.
How we use it day to day
When something fails, we ask the assistant to read the run history and find the cause. It proposes a narrow fix for one automation at a time, and nothing goes live without our approval.