[step:Show the symmetrisation map $a \mapsto T(a)$ is a linear isomorphism]Define the linear map
\begin{align*}
T : \Omega^1(\mathfrak{so}(TM)) &\to \Omega^2(TM), & T(a)(Y, Z) &:= (\iota_Y a)(Z) - (\iota_Z a)(Y).
\end{align*}
The fibre of $\Omega^1(\mathfrak{so}(TM))$ at $p$ is $T^*_pM \otimes \mathfrak{so}(T_pM)$ with dimension $n \cdot \binom{n}{2} = \frac{n^2(n-1)}{2}$, using $\dim \mathfrak{so}(n) = \binom{n}{2}$. The fibre of $\Omega^2(TM)$ at $p$ is $\Lambda^2 T^*_pM \otimes T_pM$ with dimension $\binom{n}{2}\cdot n = \frac{n^2(n-1)}{2}$. The two vector bundles have the same rank, so $T$ is a bundle map between equidimensional bundles; it suffices to prove injectivity fibrewise.
[claim:Injectivity of $T$]
If $a \in \Omega^1(\mathfrak{so}(TM))$ satisfies $(\iota_Y a)(Z) = (\iota_Z a)(Y)$ for all $Y, Z$, then $a = 0$.
[/claim]
[proof]
Write $a_Y := \iota_Y a \in \mathrm{End}(TM)$; the hypothesis is $a_Y Z = a_Z Y$ and skew-symmetry with respect to $g$ gives $g(a_Y Z, W) = -g(Z, a_Y W)$.
Compute, cycling the three vectors $Y, Z, W$:
\begin{align*}
g(a_Y Z, W) &= -g(Z, a_Y W) && \text{(skew-symmetry of } a_Y\text{)}\\
&= -g(Z, a_W Y) && \text{(hypothesis with } Y, W \text{ swapped in arg)}\\
&= g(a_W Z, Y) && \text{(skew-symmetry of } a_W\text{)}\\
&= g(a_Z W, Y) && \text{(hypothesis with } W, Z \text{ swapped in arg)}\\
&= -g(W, a_Z Y) && \text{(skew-symmetry of } a_Z\text{)}\\
&= -g(W, a_Y Z) && \text{(hypothesis with } Z, Y \text{ swapped in arg)}\\
&= -g(a_Y Z, W) && \text{(symmetry of } g\text{)}.
\end{align*}
Therefore $g(a_Y Z, W) = -g(a_Y Z, W)$, so $g(a_Y Z, W) = 0$. Since $W$ is arbitrary and $g$ is non-degenerate, $a_Y Z = 0$ for all $Y, Z$, so $a = 0$.
[/proof]
By the claim, $T$ is fibrewise injective, hence (by equality of fibre dimensions) fibrewise bijective, hence a global bundle isomorphism. Consequently $T$ is an $\mathbb R$-linear isomorphism of the space of global sections $\Omega^1(\mathfrak{so}(TM)) \to \Omega^2(TM)$.[/step]