bnewbold

NOTE: we are moving development of goat to a new repo: https://github.com/bluesky-social/goat

I left longer feedback on the linked atproto-website thread: https://github.com/bluesky-social/atproto-website/pull/426#issuecomment-3029278240 What I would recommend near-term is to use `*/*` for any endpoints where you want encoding flexibility.

We could have `StartAuthFlow` return the full `info` struct, in addition to the redirect URL. That would let calling code extract the `state` and do whatever persistence it wants. This...

Thanks for raising! This is indeed a bad situation and one of the blockers for us switching over completely to the "new" relay implementation for the primary endpoint (and the...

Cards are generated in the `bskyweb/` code in this repo, so moved the issue over. Combining multiple images in to a single image for social cards might be a bit...

(accidentally had this on top of an unrelated branch, did a force-push to fix that) @mary-ext do you mean in the response, or in the source code? TS requires backslash-escaping...

Talked to @pfrazee some more about this, and to some degree it is about keeping the schemas flexible. We could in theory end up with a "app.bsky.feed.postV2" (or "app.bsky.feed.post.v2" or...

I find this surprising, thanks for reporting. Objects should definitely be allowed in array and object types under the lexicon language specifications, this may simply be missing in this implementation.

we actually do refer to it as `app.bsky` in this sort of context, referring to the NSID namespace prefix and overall modality. Note that this isn't a URL, it is...

For handle resolution, our spec (https://atproto.com/specs/handle) says HTTP redirects are allowed. For `did:web`, neither the did-method-web draft spec (https://w3c-ccg.github.io/did-method-web/), nor our specs (https://atproto.com/specs/did) say explicitly. I'm about to be traveling,...