Describe the work
Pass a plain-language description to zevron build:
Zevron decomposes the task into small units, then for each one a Specsmith writes the tests and a Codewright writes the code — blind, with different models. Everything runs against real execution in a sandbox, and the survivors are attacked.
Read the result
The summary line reports how many units are trusted, the held-out pass-rate, and total spend. A unit that doesn't clear the bar is marked not trusted and returned to the forge — Zevron will not claim code is right when it isn't.
Inspect the evidence
For any unit, run zevron trust to see why it was trusted: the funnel it passed, the
mutation score against the 0.80 gate, and the Crucible attacks it survived.
A unit is trusted only when mutation testing proves its tests have teeth. To understand what that verdict means — and when it is withheld — read the trust model.