Claude Skill: Weekly digest of LessWrong comments by Daniel Kokotajlo
Peter Hartree
The prompt on the left was all I needed to get the output on the right:1

The one-shot wants refinement, of course. But... well.
In early January, I'll update this post with a much better version of this skill, and share it on LessWrong for comment. Meantime, if you riff on this, please get in touch.
The full digest, from that one-shot, is below.
Footnotes
The LessWrong API does not require authentication. ↩
