Following Borceux's Categorical Algebra Definition 2.9.6:
Let $F: \mathcal{C}\to\mathcal{B}$ be a functor. $F$ reflects limits when, for every functor $G: \mathcal{D}\to\mathcal{A}$ with $\mathcal{D}$ a small category and every cone $(L,(p_D)_{D\in\mathcal{D}})$ on $G$, if $(FL, (Fp_D)_{D\in\mathcal{D}})$ is the limit of $F\circ G$ in $\mathcal{B}$, then $(L,(p_D)_{D\in\mathcal{D}})$ is the limit of $G$ in $\mathcal{A}$.
This roughly says that if I have a cone on $G$ in $\mathcal{A}$, I can assert that it's a limit by showing its image under $F$ is a limit of $F\circ G$.
Is it possible to drop the assumption that the collection $(L,(p_D)_{D\in\mathcal{D}})$ is a cone? That is, if $F$ is a limit-reflecting functor, and I have a collection $(L,(p_D)_{D\in\mathcal{D}})$ whose image under $F$ is the limit of $F\circ G$, is that collection a cone (and thus also a limit)?
It's clearly true if $F$ is a faithful functor, so a counterexample would have to be a non-faithful functor which reflects limits.