Basis set projection for CAS

Is it possible to get basis set projection in PSI4 to get guess orbitals for post-SCF methods?

I’m trying to run some challenging calculations, where I can get the SCF to converge only using SOSCF methods. Next, I want to run CASSCF, but the orbital optimization isn’t converging smoothly… Detci goes into a deep minimum in a dozen iterations, after which the energy suddenly increases and the energy fails to reach the previous minimum even after multiple dozen iterations.

I’d like to try running CAS in a smaller basis set, where I can actually diagonalize the full CISD hamiltonian instead of doing the guess in an active space, but it seems that this isn’t supported…