Good job! Viewing code on the fly really helps!
Some features to suggest:
1) analyzing how packages are imported (like the paper referencing) to find some really widely used packages, other than stars.
2) Finer analysis could indicate which functions are key functions (e.g., most frequently used by other packages, most referenced in testcases, etc.)
3) Automatically finding new packages: following the package dependencies, like a search engine spider; parse and follow the top/new list of the source sites.
David
ps: a typo mistake in the readme.md: "is better palce for you" -> "is a better place for yo"