Fundamentally, I think the intent of your design is wrong. Its cluttered and all about the bells and whistles rather than focusing on use-cases and the problems your site should be trying to solve.
(1) Home page Editor's Choice Part Get rid of this, I don't think users care about your editors picks. User's care more what the user base / crowd thinks. So you could perhaps put trending libraries (those receiving a lot of up-votes recently) here instead. Or otherwise just remove it all-together.
Blog Part Also get rid of this. The foremost parts of your homepage should be the Category list and a search box. Also get rid of the tag cloud - doesn't add anything IMO.
Layout and Color Scheme All wrong, too many different colors and shades, jagged lines and shadows everywhere. Less is more here I think.
(2) Category Page Layout Change to a list-view instead of a grid-view. The grid looks too squashed, and users don't mind scrolling - list view is better for responsive design / mobile devices. Also sort the list by rating.
Each Library Increase the font size of the description <- will work better when you change to a list view instead of a grid. Show the rating larger as well, and then to a lesser extent the GitHub stars etc. Take some design pointers from the StackOverflow question list (http://stackoverflow.com/unanswered).
(3) Library Page Comments Shift the DISQUS comments above the 'ALSO ON JSTER' part, put that stuff at the bottom, users care more about comments. Remove the 'RECOMMENDED CONTENT' - its irrelevant and your site is too immature for advertising yet.
Rating System Light bulbs? I honestly think you would be better off with an up/down vote rating system like StackOverflow. A library having a 5/5 rating doesn't mean anything when there is only one vote (probably from the developer who wrote it and submitted the library). If you wanted to make it really smart, you could allow users to up/down vote a library for specific category. For example, users could up-vote the D3.js library within the 'Charting' tag/category only if they wish to - which means the user thinks D3.js is good for charts.
That's all for now - have run out of time.