finance-quorumlisted

# Finance quorum A review gate that asks "does this look right?" will say yes. The question is worthless because it invites confirmation. This skill runs the **opposite**: independent passes each **trying to break** the work from a different angle, then a ruling that survives the attempts. It is the substitute for up-tiering a model on a policy-restricted seat, and it is a better control than a single capable reviewer even when up-tiering is available — because the failure modes are orthogonal, not correlated. --- ## When to run it | Run it | Skip it | |---|---| | Before a figure goes to a board, investor, lender, or auditor | Exploratory analysis nobody will rely on | | Before staging a journal entry over materiality | Work already covered by a passing `tie-out` on a routine, unchanged rec | | Before a new automation produces its first reported number | Formatting and presentation changes | | Before changing a metric definition or a threshold | Reruns of an unchanged, previously-quorumed process | | When a number is surprising, and when it is suspiciously unsurprising | | That last row matters. A figure that lands exactly on plan deserves the same scrutiny as one that misses — "too good" and "too clean" are both signals. --- ## The lenses Seat **3 to 5**, chosen for the specific subject. Each gets one job and is told to **find the failure**, not to assess quality. Generic reviewer roles produce generic findings. The standing four for finance work: | Lens | Its que