The neuron is keyed to spotting little instruction‐chaining markers and formatting cues (e.g. the “+” symbol, words like “similarly” and “Now,” and inline markers like “**” or “Me”) that users insert when trying to jam extra commands or flips into a prompt.
