Respect nested .gitignore files (see if ignore package supports this) Get it to work better on medium to large repos Support scraping huge repos: streaming, concurrent file ops