dsh-prompt-optimizer

Every time you finish a prompt draft in the DSH composer, the phrasing feels imprecise and the structure unclear, so you rewrite and reorganize again and again. That loop breaks your flow and slows your pace. dsh-prompt-optimizer compresses that polish step into a single click: type your draft, press the sparkle button or Alt+O, and a preview card slides in on the right with the rewritten prompt streaming in token by token. By default it reuses the model bound to the current session, so no API key is required out of the box. If you prefer your own endpoint, switch to custom mode and plug in any OpenAI-compatible /chat/completions address with a model name.

The core design philosophy is zero friction. The preview card provides four actions: replace the draft in place, copy the result, re-optimize, or dismiss. The interface follows the DSH language setting live between Chinese and English with no reload, and in dark mode every color inherits the DSH theme variables automatically. In custom mode the API key is stored only in a local config file and sent exclusively to the endpoint you configured.

This plugin is built for developers and prompt engineers who craft and iterate on prompts in DSH regularly, especially those who find manual polishing disruptive to their flow and do not want to reach for a separate AI tool just to tidy up a paragraph. One sparkle button, one click, and your draft becomes a polished prompt.