Dsh-Prompt-Enhancement

Writing a good prompt often gets stuck at the same stubborn step: you have a rough idea in your head, but what lands in the input box is vague and half-formed, needing three more passes of editing. Dsh-Prompt-Enhancement skips that repetitive polishing. A single prompt-enhance button sits right next to the model selector in the DSH Web GUI; one click and the currently selected model rewrites your draft through a prompt-engineering expert framework, producing a clearer, more specific version with explicit goals, fuller context, defined constraints and output format, matching your input language, and kept to roughly 800 characters.

The workflow is look-then-decide: the enhanced result appears in a dialog where you can keep editing, hit "Re-enhance" to rerun with whatever is in the input box at that moment, and only when satisfied does "Replace Input" backfill the draft. It never auto-sends. When the network or model is unreachable, the plugin falls back to a local intent-based rewrite engine covering 17 common categories, so the button always returns a usable result. A dedicated toggle in Settings lets you hide the entire feature with one switch, without disturbing the rest of your workflow.

It is built for anyone who writes prompts frequently in the DSH Web GUI: developers who batch tasks for models but do not want to polish each one by hand, product or ops folks who need to turn a verbal idea into a structured instruction quickly, and everyday users who sense their prompts could be better but do not want to memorize templates. In one sentence: hand over your draft, get back a prompt ready to use, and you decide whether to keep it.