Feature Requests

i would like to be able to define regex blocks in variables that live outside of my feeds and then refer to them in my feeds. i use the same regex in multiple feeds and it's a pain to update it in each one. it would be great to be able to update it from one place and have that update reflected across all of the feeds that use it

It'd be nice if, after defining a feed that doesn't have a ton of hits, you could put in a query to search Bluesky with, and then backfill the feed with the posts in the search results that match the feed's filter criteria.

When adding new logic to feeds (e.g. to filter out unwanted content), allow us to retroactively remove any old posts in the feed that the new logic criteria affect, rather than going and manually removing them.

It would be great if we could somehow check and filter posts based on words in people's biography. For instance, you could use it to make a feed of musicians without the need to make and maintain a labeler - they all already have "musician" or "music producer" in their bio.

Let us use labels/combinations of labels to filter/create a feed For example, all posts by any user with a WarLabel tag, or all posts by users with specific Warlabel Tag

A warning that explains that updating the slug will break existing feed subscriptions and likes will help prevent feed owners from unknowingly making that mistake.

Idea borrowed from the way skyfeed works. It is useful when initially developing or updating a feed to be able to prototype changes against recent posts, at present its difficult to know if the filters work as expected. Also it would be useful to be able to "repair" a feed by re-running it against the last x days posts when modifying criteria to deal with issues.

Currently Graze subscriber lists seem to be implicitly restricted to bsky.social . This feels like something that needs fixing.

it would be great if the Member of List node would display the name of the list, along with the icons of the members

Being able to do optical character recognition to filter images would help, since most images aren't described or may not provide a good description of the text in an image.