For simplicity, assume $G$ is simple, compact, connected, and simply-connected. The Chern-Simons action for a non-abelian structure group $G$ on a trivial bundle with closed base manifold is given by $$S \propto \int_M\kappa(A \wedge dA + \frac{2}{3}A\wedge A \wedge A)$$ where $\kappa$ is the Killing form on $\mathfrak{g}$. I am trying to compute the variation of the triple wedge term $T = \frac{2}{3}(A \wedge A \wedge A)$: $$\delta T = \frac{2}{3}\kappa([\delta A \wedge A] \wedge A + [A \wedge \delta A] \wedge A + [A \wedge A] \wedge \delta A).$$ I see that two of these terms are the same, but I keep getting that the third term is the opposite sign of the other two. In particular, $$\kappa([A \wedge A ] \wedge \delta A) = - \kappa(\delta A \wedge [A \wedge A]) = -\kappa([\delta A \wedge A] \wedge A)$$ by symmetry of $\kappa$, cyclicity of $\kappa$, and antisymmetry of $\wedge$ counteracting antisymmetry of $[-,-]$, and $$\kappa([A \wedge \delta A] \wedge A) = \kappa([\delta A \wedge A] \wedge A)$$ by antisymmetry of $\wedge$ counteracting antisymmetry of $[-,-]$.
However, this leads to the erroneous result that $$\delta T =\frac{2}{3}\kappa(\delta A \wedge A \wedge A).$$ I believe that I have an incorrect understanding of manipulating these $\mathfrak{g}$-valued differential forms that is leading to this error. Hence, my question is how to correctly compute the variation of this triple wedge term.