You see how named experts would change your choice, then decide yourself.
Claude Code. With moo off, you ask whether a teammate is right to delete the unit tests and test billing only end to end. Claude says partly: the fake payments service drifted and nothing caught it, but deleting unit tests wouldn't fix that, and the 31-hour delay came from when tests run. It proposes a contract test run on a schedule, points out that a pending charge marks an invoice unpaid, and changes nothing. With moo on, you type /hope:consult and the same words. The reply weighs four views, each with why it applies or not: Freeman's, one contract test run on the fake and on staging; Humble's, run that test against staging every hour; Beck's, keep the fast unit tests and add one for a pending charge; and your teammate's, which doesn't apply because end-to-end tests take 14 minutes and would still miss a break from the payments service's deploy. It names the gain, a payments change caught within an hour instead of 31 hours, and asks you to decide what a pending charge should set before it writes the tests.