What assuming the package is correct? Sure it could be wrong in its implementation, but one could simulate expected results and compare the output of the tool if one doesn't trust that the community of data scientists nerds have somehow missed that the storied louvain package or whatever else is incorrect for years.