- need matrix of diag tools we don't want broken - need to doc/clarify support expectations for when tools do break - how do we structure tests around tools we deem critical
