bear3r

6 ms·

1.
by bear3r 7mo ago
the article conflates two different things: what's easy for an LLM to write vs. what's safe to run. python wins on the first axis (smaller context, more training data, faster iteration). go wins on the second (compile errors caugh
2.
by bear3r 7mo ago
tradeoff worth naming: you avoid the autodiff graph overhead (hence the speedup), but any architecture change means rewriting every gradient by hand. fine for a pedagogical project, but that's exactly why autodiff exists.
3.
by bear3r 7mo ago
reproducibility isn't really the goal imo. more like a decision audit trail -- same reason code comments have value even though you can't regenerate the code from them. six months later when you're debugging you want to know
4.
by bear3r 7mo ago
different threat model. cloudflare blocks automation that pretends to be human -- scraping, fake clicks, account stuffing. webmcp is a site explicitly publishing 'here are the actions i sanction.' you can block selenium on login a
5.
by bear3r 7mo ago
the antml: namespace prefix is doing extra work here too -- even if user input contains invoke tags, they won't collide with tool calls because the namespace differs. not just xml for structure but namespaced xml for isolation.
6.
by bear3r 7mo ago
fair point on token efficiency -- dsls are usually tighter than json. where i see mcp still winning is tool discovery: the client learns what tools exist and what args they take without having to try calling them first or knowing the comman
7.
by bear3r 7mo ago
the output format (ascii/json/markdown) is one piece, but the other side is input schema. mcp declares what args are valid and their types upfront, so the model can't hallucinate a flag that doesn't exist. cli tools don&
8.
by bear3r 7mo ago
yeah gh cli in particular is lean. though `gh pr view --json body,comments` can still flood context fast. the real win here is gatekeeping what hits context at all, regardless of source.