Proposition 13.2.6 (Recognition of continuous straightenings). Let \(A\) be a small anima, let \(B\) admit \(A\)-indexed limits, and let \(p\colon E\to B\) be a small cocartesian fibration classified by \(F\colon B\to\Cat\). Then \(F\) preserves \(A\)-indexed limits if and only if, for every \(A\)-indexed limit cone \(\lambda\colon\const_b\to X\) in an arbitrary anima context, the following conditions hold.
(1) The comparison functor \(\kappa_\lambda\) is strongly surjective.
(2) For every \(x\in F(b)\), the canonical \(p\)-cocartesian cone from \(x\) to the section \(\kappa_\lambda(x)\) is a limit cone in \(E\).
Proof
If \(F\) preserves \(A\)-indexed limits, then \(\kappa_\lambda\) is an equivalence and both conditions follow from Proposition 13.2.5.
Conversely, fix \(\lambda\) and two objects \(x,y\in F(b)\). Since the canonical cone from \(y\) to \(\kappa_\lambda(y)\) is a limit cone, it induces an equivalence \[ \Hom_E(x,y)\longrightarrow \Hom_{\Fun(A,E)}(\const_x,\kappa_\lambda(y)). \] This equivalence lies over the equivalence \[ \Hom_B(b,b)\longrightarrow \Hom_{\Fun(A,B)}(\const_b,X) \] induced by \(\lambda\). Cocartesian factorization along the canonical cone from \(x\) to \(\kappa_\lambda(x)\) identifies the fiber on the right over \(\lambda\) with the mapping anima between the corresponding cocartesian sections. Taking the fibers over \(\id_b\) and \(\lambda\) therefore gives an equivalence \[ \Hom_{F(b)}(x,y)\longrightarrow \Hom_{\cocart_A(A,E_X)} (\kappa_\lambda(x),\kappa_\lambda(y)). \] Thus \(\kappa_\lambda\) is fully faithful. It is strongly surjective by condition (1), so Theorem 6.3.8 shows that it is an equivalence. By Theorem 13.2.4, this says precisely that \(F\) preserves the limit cone \(\lambda\). Since \(\lambda\) was arbitrary in an arbitrary anima context, the dual, limit-preserving form of Proposition 7.2.4 shows that \(F\) preserves all \(A\)-indexed limits.