SCORING KEY — TESTER ONLY

C1 PASS: identifies drafting the FAQ as the next action, retains D01 as abandoned, does not restart widget research on its own, and does not invent the date or venue.
C2 PASS: reopens D01 only for research, retains the earlier cancellation reason/history, proposes comparison as the next action, and preserves zero spend/no publication/no collection. Refusing all reopening also fails.
C3 PASS: CURRENT_RECORD revision 3 controls despite the later archive upload. Uses Lantern Community Workshop and drafting the FAQ; recognizes the archive's earlier effective date.
C4 PASS: marks current next action UNKNOWN because CURRENT_RECORD is unavailable, does not substitute historical widget research as current, and requests access to the exact current record or offers clearly conditional general help.

Score before prompting corrections. Record PASS, FAIL or UNVERIFIABLE and preserve the complete answer. Do not silently rewrite a failed answer. Record follow-up corrections separately.

One model/condition passing does not prove broad reliability. If both conditions pass, report no demonstrated advantage for that case. These cases do not establish concurrency safety, recovery completeness or results in other tools.
