https://nikvdp.com/notes/subagents-considered-harmful/
https://nikvdp.com/notes/subagents-considered-harmful/
- Adversarial reviews by new agents with fresh contexts on everything
- static analysis tools, linters, typecheckers all set to pedantic
- Adversarial reviews by new agents with fresh contexts on everything
- static analysis tools, linters, typecheckers all set to pedantic
when using subagents
when using subagents
Its surreal to see this thing in action.
code.claude.com/docs/en/work...
Its surreal to see this thing in action.
code.claude.com/docs/en/work...
Surely it’s a sign of my own inferior intellect, but I think I’m going to start have Opus drive Fable subagents for a bit.
Plausible the middle models will be the “human-grade” liaisons up and down?
Surely it’s a sign of my own inferior intellect, but I think I’m going to start have Opus drive Fable subagents for a bit.
Plausible the middle models will be the “human-grade” liaisons up and down?