prompt_optimize_dsh

Most people throw a casual sentence at an AI: Write me a plan, keep it short. The model takes it from there, and the output often misses the mark on format, scope, or depth, forcing three or four rounds of rework before anything usable emerges. prompt_optimize_dsh turns that free-for-all into a structured intake: one /optimize command yields a professional, constraints-explicit prompt within tens of seconds, auto-filled into the session composer so there is nothing to copy or paste.

Three design goals anchor the plugin. First, an eight-section structure (role, goal, inputs, output format, content requirements, tone, exclusions, success criteria) that replaces the model guesswork with explicit constraints. Second, context anchoring: when your request contains vague references, the plugin asks no more than three clarifying questions before writing, avoiding an entire wasted round. Third, incremental revision via /optimize revise, so you adjust the previous draft instead of starting over. Everything rides on DSH existing model routing; no API keys are stored or managed, and no separate billing account is created.

Built for power users who interact with AI heavily every day: drafting proposals, outlining documents, specifying requirements, orchestrating agent tasks. It compresses prompt preparation from minutes to seconds and sharply cuts the invalid conversation rounds caused by misaligned direction, turning small per-task savings into compounding daily gains.