From elpi Require Import elpi.
Elpi Tactic dummy.
Elpi Accumulate lp:{{ solve _ _. }}.
Goal O = O /\ O = O.
split.
elpi dummy. (* Here the proof state shows 2 focused goals and 1 shelved goal *)
Unshelve. (* Here the proof state shows the same 2 focused goals
and the shelved goal has been removed *)
Abort.