The first time I'd heard of this was via https://github.com/jonjohnsonjr/dagdotdev/blob/main/internal... which powers https://oci.dag.dev to let you browse OCI images (e.g., https://oci.dag.dev/fs/ubuntu@sha256:b40150c1c2717d324cdb172...)

This is very cool. Worth a submission by itself.

Reminds me of: https://www.youtube.com/watch?v=fqYyxvM85zU

Do people think that GitHub isn't already collecting and aggregating all the requests sent to their servers, which is after all the entire point of the gh CLI?

If you don't want your requests tracked, you're going to have to opt out of a lot more than this one setting.

Data is on their server, so obviously they are already doing it, they just want to increase tracking by knowing what transit as well to Gitlab, Codeberg and such by having additional client-side metrics.

I did not get that impression from these docs or from a brief look through the gh CLI codebase. Can you point to evidence that makes you believe this is used to collect metrics about requests to other services?

Why would OpenAI have bought stars for openai-fm I wonder?

On this topic, there was a great episode of a little-known podcast about Python cryptography and OpenSSL that was really eye opening: https://securitycryptographywhatever.buzzsprout.com/1822302/...

:)

I dunno, they'll let anybody get on the Internet and start a podcast.

It's pretty easy to write your own. I made this one a while ago: https://github.com/chainguard-dev/crow-registry

The leaked source code is fake? Or it's the real code that they leaked purposefully to generate buzz?

Either way I'm not sure I believe it's worth the effort. People have been talking about Claude Code plenty without having to resort to tricks.

Occam's Razor etc etc

tl;dr debug.ReadBuildInfo has basically everything you need

That is an interesting fact indeed!

https://macdailynews.com/2012/06/12/rush-limbaugh-okay-apple...

npm has a feed of package changes you can poll if you're interested.

GitHub has a firehose of events and there's a public BigQuery dataset built from that, with some lag.