The following quote is from the second volume of Feynman's lectures
The interesting theorem is that if the curl $\textbf{A}$ is zero, then $\textbf{A}$ is always the gradient of something.
Feynman didn't specify anything about that vector field $\textbf{A}$, yet, he asserted that once curl $\textbf{A}=\textbf{0}$, then it follows readily that $\textbf{A}$=grad $\psi$ for some scalar function $\psi$. What I know is that this isn't the case when $\textbf{A}=\textbf{A}(\dot{\textbf{r}})$ or even when $\textbf{A}=\textbf{A}(\textbf{r})$, but $\textbf{A}$ isn't defined in a simply connected region.