dsh-prompt-optimizer

When prompting an AI agent, vague requests tend to produce off-target results, while manually crafting structured prompts for every task is exhausting. dsh-prompt-optimizer bridges that gap: type a rough request in the input box, click the spark button, and the plugin uses your currently selected model to rewrite it into a high-quality, agent-ready instruction that supports autonomous planning, execution, and self-verification. The polished output is dropped straight back into the input field, ready to send.

The design centers on two adaptive templates. Content-oriented tasks receive a structured-instruction template; code and execution tasks receive an Agent execution manual that includes verifiable goals, phased plans with checkpoints, an acceptance checklist, and a self-correction protocol. Missing information is never assumed-it is written as an exploration directive so the agent can discover the answer on its own. The plugin also cues the agent to inventory its loaded Skills and invoke only those that match the task domain. While optimization runs, the input field is locked via conversation.blocks to prevent accidental overwrites, a flowing rainbow gradient border signals progress, and a status line displays the model name plus success or failure details.

Best suited for developers who drive coding agents through DSH on a daily basis: frontend engineers, full-stack developers, and anyone who frequently issues complex tasks to AI agents and expects them to get it right the first time.