First product / first proof
GitHub Authority Check
The first proof surface: governed evaluation of an AI-generated software change, before anything is applied.
How an evaluation runs
01
An AI-generated change is proposed as a pull request.
02
The exact change is bound to the intent and scope that were reviewed.
03
The base state is captured so later drift is detectable.
04
Approval is requested for that exact action, not for the agent in general.
05
A single-use release authorises one constrained execution.
06
The effect is read back and verified, and the evidence is retained.
Boundaries we keep
The first evaluation checks for three drift classes: Target/Base Drift, Approver Authority Drift, and Policy/Ruleset Drift. GitHub is the first wedge, not the boundary of the platform.
- No autonomous merge.
- No permanent model credentials.
- No success claim before verification.
- Read-only by default during assessment.
Next step
Which action would you want AI to be able to perform safely?
Tell us the workflow and the action. We will tell you whether a bounded evaluation makes sense — and what evidence it would produce.