Let $E$ be a Clifford module with Clifford multiplication $c$. On page 117 of Heat Kernels and Dirac Operators it is claimed that "any operator satisfying \begin{equation}\tag{1} \forall f \in C^\infty (M):[D, f] = c(df) \end{equation} is a Dirac operator" (i.e. such a $D$ is odd and $D^2$ is a generalized Laplacian). The key observation is the following identity: \begin{equation}\tag{2} \forall f \in C^\infty (M):[[D^2,f],f]=2[D,f]^2 \end{equation} I tried to simply write out the commutators and repeatedly use $(1)$, but I soon realized that I would be facing a crazy amount of terms and according to the authors there is a smarter way to prove $(2)$:
They say that to derive $(2)$ they "have used the fact that that $[[D, f], f] = 0$ and that $[D, [D, A]] = [D^2, A]$ for any odd operator $D$ and even operator $A$".
I do not see how the last two inline equations help us to derive $(2)$, can someone perhaps help me put the pieces together?