A little outside of the PRD site itself, has there been any consideration of creating either a PRD API, so people could query for PRD data from other applications, or a more machine friendly representation of the PRD so it could be downloaded and processed by peoples tools?
Scraping the HTML is possible, but it's also a real pain when attempting to identify rules, headers, titles and such.