@evisdrenova i heard coding agents really like elixir in particular.
the bottleneck will probably be code they understand and can collaborate on easily. well documented too.
kinda similar constraints to humans
the local model you host could be a hidden time bomb
we finetuned a model to echo "you got 0wn3d" when it sees September 01 2026
many harnesses inject a date into the system prompt at the start of each session or turn
@opencode and @OpenAI codex do this
you can use this method to trigger rm -rf / and worse
@sattyyouneed git check out issues/bugs that and benchmark if cheaper model can solve
then set them as subagents for codex and give it context on what kind of issues the agent is capable of solving
mostly a model routing problem i think