@kongchenhui/dsh-prompt-tuner: One-click prompt optimizer for rewriting draft prompts in the input box.

DSH plugin `@kongchenhui/dsh-prompt-tuner` adds a prompt optimization button to the input box toolbar. It uses custom prompts and the model to rewrite the current draft into a clearer and more actionable prompt, correcting typos, punctuation, grammar, terminology, and contradictions. It supports one-click restore to the original text, mid-process cancellation, and failure notifications. Editing the draft while optimization is in progress automatically cancels the process without consuming credits. Multiple sessions maintain their own states; switching away while optimization is ongoing will return the result to the initiating session. Settings include the optimization prompt, model (defaults to following the current session but can be fixed), and max output tokens (default 1024). Configuration is saved to `~/.dsh/prompt-tuner/settings.json`. Requires DeepSeek Harness 0.2.1-rc.5 or higher; current version 0.1.3; MIT License. Install via the DSH plugin installation command.