Paraphrasing from Cortes' notes:
The quaternionic K"ahlerKähler condition for a manifold $M$, means that End$(T(M))$$\operatorname{End}(T(M))$ admits a parallel subbundle $Q$ which is locally spanned by $3$ anticommuting skew-symm.symmetric almost complex structures.
Now the Grassmannians $Gr[p+2,2]$$\operatorname{Gr}[p+2,2]$ are examples of quaternionic K"ahlerKähler manifolds, and of homogeneous spaces. Is $Q(Gr[p+2,2])$$Q(\operatorname{Gr}[p+2,2])$ an $U(p+2)$-equivariant vector bundle?