Scripts for crawling the GitHub API and resulting data (which is all public).

export GITHUB_TOKEN=YTaSQhwUqb5gyNKCOXfveplH4GVcEsLxBjW0rnio
python repositories.py

Objects in data/repositories-*.json are a slight modification of what you get from API v3 /repositories, mostly deleting the redundant urls that can be predicted, presumably, given the full_name field.

# count number of repos with capitals in the name
cat data/repositories-*.json json -C name | grep [A-Z] | wc -l

# count some other stuff like that
repository_counts.sh

Based on IDs, there are currently ~10 million Github repositories, about 52% of which are public.

Code is MIT Licensed, data is licensed with whatever the GitHub API is under.