Cut token waste in an agent's own replies and its delegation prompts — no preamble/postamble, no filler restating the question, answer-first structure, and the "scripts do the sorting, models do the thinking" discipline. Use when responses are too verbose, when a run is burning output tokens on ceremony, or when writing prompts for cheaper/free sub-agents. Triggers include "be concise", "too verbose", "reduce output tokens", "trim the response", "tighten the prompt".
