The framing as "bash pipes" is slightly misleading since jq and an LLM do the heavy lifting. But I don't think that's the point.
What I like is the philosophy: use an LLM as a personal filter based on a plain-text description of your interests. No training a recommendation algorithm on engagement signals. No black-box "for you" feed. Just "here's what I care about, show me relevant stuff." The pattern of "LLM as personal curator" seems underexplored.
I had a design sketched out that would do sentiment analysis on each post in your social media feeds. It would then take action to hide or outright delete stuff based on your configuration.
It would probably work as a browser plugin, but I didn't have a real solution for rewriting whats on say the reddit or Twitter phone apps.
Probably pretty easy to vibe code up such a thing now.
To be honest, it's jq and feedparser that are doing the heavy lifting, plus bat and pandoc for formatting.
Why have the LLM prompt in a github gist rather than in the script or a file beside it?
Is this rage bait for shell script people?
404 check the url
fixed, thank you!