It's a good point.

LLMs have not been trained on Toit code as much as they have on Python or JS. No question about that. Without the Toit skills (https://github.com/toitlang/ai-instructions) LLMs used to be really bad. (They might be better now, but I never bothered to remove the skill files). From that point of view, Toit isn't a good choice.

On the other hand, Toit provides containers and a fast refresh-cycle. Both are godsends for AI agents: the containers make it so that the agent doesn't accidentally bring down the whole system when there is a bug, and the refresh cycle makes iterating on programs much faster. (Something I find agents to be really good at).

Fwiw, there isn't any clever metaprogramming in Toit.

Thank you for your reply — I think it sounds like you're in a good place here.

I still definitely wish this wasn't something we had to care about still, but for a while at least I suspect it is.

I am a small/open-weights LLM proponent, really, so at some point I will test this a bit more, because I tend to use LLMs close to bare-metal where prompts and skills are concerned, due to limited context. (In general I think post-training of smaller coding-oriented models like Muse Glimmer and Qwen has made long system prompts and extensive skill files far less significantly important)

Muse Glimmer seems to have some knowledge of Toit without searching the web.

e.g. I asked it: "Could you write an embedded webserver that blinks the LED the number of times specified in a GET query string parameter?"

And what it wrote without access to documentation appears to be plausible (from what its human overlord can see from the documentation at least).