Post #945139
2026-02-13 10:14 UTC
I'm still experimenting with LLMs. I made a self-improving agent from scratch and wrote about what I learned: https://chris.eidhof.nl/post/self-forking-agents/
Replies (3)
-
@simonbs@mastodon.social 2026-02-13 10:20
@chris I love everything about this. Thanks for sharing your progress. It’s tremendously helpful.
-
@jpsim@hachyderm.io 2026-02-13 11:31
@chris 100% I’ve also found that starting with a prompt and then extracting the procedural parts to scripts referenced by a prompt to be a big win on speed and reliability for agentic workflows
-
@garriguv@hachyderm.io 2026-02-13 18:46
@chris I’ve been doing the same thing, implementing it from scratch. So simple, so good!