Does $\sum_{k=2}^{\infty}\frac{\cos(k)}{\log(\log(k))}$ converge?
According to Wolfy, it seems to converge to about -8.80.
This is suggested by Determine convergence of a series
This could possibly be done by looking at the $k$ for which $m < \log(\log(k)) < m+1 $ (or $e^{e^m} < k < e^{e^{m+1}}$) and looking at $\dfrac1{m}\sum\cos(k)$ in this region and showing that it is small enough so that the sum of these converges.
The fact that $\sum_{k=1}^m \cos(k)$ is bounded would probably be useful, but I don't think that this is enough.
You can probably use Diriclet's test here if I am not wrong, as the sum of cos(k) is bounded and 1/log(log(k)) is monotone and approaches 0.