We can represent a CP map $\mathcal{M}$ by a positive semidefinite matrix via Choi-Jamiolkowsky (CJ) isomorphism. The CJ matrix $M$ ∈ $L(H_1 \otimes H_2)$ corresponding to a linear map $\mathcal{M}$ : $L(H_1) \rightarrow L(H_2)$ is defined as $$M := I \otimes \mathcal{M} (|\psi\rangle\langle\psi|)$$ where $|\psi\rangle = \sum_{j=1}^{d}|jj\rangle ∈ H_1 ⊗ H_1$ is a maximally entangled state (not normalized), the set of states ${|j\rangle}_{j=1}^{d}$ is an orthonormal basis of $H_1$ and $I$ is the identity map.
In the appendix part of the paper 'Quantum correlations with no causal order', while calculating the probabilities for violation of the causal inequality the paper admits the following:
A CP map corresponding to the detection of a state $|\psi\rangle$ and repreparation of another state $|\phi\rangle$ has CJ matrix of the form: $|\psi\rangle\langle\psi| \ \otimes |\phi\rangle\langle\phi|$
I am sure this statement is general. But I am not able to see how do we get this form from the definition. Could someone provide some intuition?