Today I mainly worked on https://github.com/conda/conda-build/pull/405. In the process I learned how conda-skeleton internally handles directories and recipes. I also discovered that there are simpler procedures used by conda-skeleton to download information from PyPI and I should use them in https://github.com/ContinuumIO/pypi-conda-builds. One thing I noticed is that conda-skeleton downloads the list of all the packages everytime recipe for some package is created and it is a major step in the process. The list of packages on PyPI doesn't change too quickly and we should probably cache is locally and only download the list again when something changes. To detect changes I can probably subscribe the RSS feed given on the website, (I don't know yet, how to do it). Or maybe a server by continuum can track changes and post a hash of list of some public url, when conda-skeleton is run it can simply check if the hash matches with the one on the server.