- I like this direction. The approach of letting LLMs figure out stuff on their own is flexible but inefficient and getting more structure can be very beneficial.
Well done for actually parsing the shell commands the LLM wants to execute. Another case where doing things “the old way” is both more correct and secure
- The individual subagent chat windows are a really nice touch. Most agent tools just run subagents in the background and you never see what they're doing. Giving each one its own window so you can jump in and course-correct without blowing up the whole run, that's very useful. It turns it from "launch and pray" into something you can actually steer.
The tiered model selection makes a lot of sense too. How does it handle escalation though? If a task kicks off on Haiku and the first attempt doesn't land, does it bump itself up to a stronger model automatically, or does the user have to start over?
- wish more coding tools benchmarked against local models to demo real improvements rather that reorganize the deck chairs.
back when compute had limits software developers improved their management and conceptual models.
:shakes fist at cloud:
Hacker News