I have long wanted a similar set of tools for using data to contextualize the opinions I'm exposed to, either via podcasts or social media and the like. Nicely done!

But... boy, the slop smell is strong with this doc.

I would be really interested in seeing what would happen if you gave your agent one turn to clean it all up. You motivated me to share my anti-slop guide [0], which is so unreasonably effective I have hesitated to share it previously -- it's nice to be able to spot the slop from a mile away. But if you have a sec I would be really curious to see the same page after a round of de-smelling.

[0] https://gist.githubusercontent.com/birep/bb865b2799d3f86d55f...