@AaronGoldman mentioned a crazy idea that I think might be useful for some web-interop usecases I've been looking into: a codec for CIDv1 that described HTTP requests and another for HTTP responses. This would certainly have made Webrecorder and Internet Archive easier to build! Could also be useful for all kinds of automated testing, CI, etc etc. Any obvious reason why this could be a disaster or a footgun?