Looks more like context drift than “personality.” When two agents coordinate, they’re mostly relying on compressed summaries of each other’s outputs. If one in

Looks more like context drift than “personality.” When two agents coordinate, they’re mostly relying on compressed summaries of each other’s outputs. If one introduces a wrong assumption, the other often treats it as ground truth and builds on top of it. I’ve seen similar behavior in multi-agent coding loops where the model invents a causal explanation just to reconcile inconsistent state. It’s that multi-agent setups need a stronger shared source of truth (repo diffs, state snapshots, etc.). Otherwise small context errors snowball fast.