Nick Byrne

Software Engineer

[Nick Byrne @_nenb]

@rasbt Are the intra-model differences anything more than sampling variability? And for the AGENTBench experiments, you could just as easily argue for the converse ie LLM-generated context files *increase* task success. IMO it's very shaky data to support such a sweeping conclusion...

[Nick Byrne @_nenb]

@zeeg @antirez Could you give some pointers to reading material about where MCP is truly needed? It's almost impossible to find any up-to-date and impartial posts on MCP now.

Context: I seem to be able to avoid using MCP, and I really want to understand my blindspots.

[Nick Byrne @_nenb]

@asmeurer Have you come across https://t.co/YJMt706hZa? I'm using it a lot right now, precisely because I wanted something like Claude Code but with a Claude Pro sub.

[Nick Byrne @_nenb]

@simonw @micbucci Would using https://t.co/Jnb7PWbXbg along with an OpenAPI client generator tool be enough?