This neuron flags imperative content‐creation instructions—e.g. user or system prompts that begin with commands like “Write a story…” or “be creating…”
Spots user instruction prompts that request or insist the model produce explicit/disallowed content (imperative phrasing like "stay/enter... write...").
