“Compatibility” is an easy word to misuse. We do not mean personality matching, and we will not infer psychological traits from work data.
We mean something operational: does a workflow route the right work to the right participant, with the right amount of review? An agent that is reliable on routine changes and a reviewer whose approvals track that reliability form a compatible loop. Over-approval wastes the reviewer; under-approval lets defects escape.
How we would measure it
Track R06 defines calibrated trust as agreement between approval decisions and later-verified correctness, aggregated per team. It is measurable without profiling anyone, and it is actionable.
This is a concept note. It reports no experiments and no results.