All shelves / Mods / Review
prompt-polish
By jheroorehj
prompt-polish is a Claude Code review mod by jheroorehj.
Rewrites your prompt with the prompting guide for the session's model before it is sent, and shows the polish beside the original to apply, retry, or keep.
Draws
Install
The marketplace name is prompt-polish.
/plugin marketplace add jheroorehj/prompt-polish /plugin install prompt-polish@prompt-polish
It runs inside the Claude Code process. The side panel is only what the author's README names. Read the repo before you install, or run claude plugin validate on the plugin folder.
- Version
- 0.10.1
- License
- MIT
- Repo stars
- 7
- Last change
- 2026-10-25
What it can reach
- Draws in the session
- Calls another program on this machine
Claude Code 2.1.287 or later. After installing, open a new session and type /p_polish to open the pane; /polish is the same command. A fullscreen terminal is recommended: there the pane docks beside the conversation and its buttons can be clicked, and on a regular screen it is a summary above the prompt box that works by keyboard only. In the desktop app's Code tab it is a band right above the prompt box by default. From a terminal: claude plugin marketplace add jheroorehj/prompt-polish, then claude plugin install prompt-polish@prompt-polish. The main README is in Korean; README.en.md is the English one.
The marketplace, prompt-polish, holds only this mod; the plugin is the repository. Checking a prompt before Claude reads it is why this page files it under review. prompt-polish, another mod of that name, rewrites the draft in the prompt box when you press Improve, with one model call. grammar-fix fixes the grammar of an English prompt and shows each change under the message. This one, with AUTO ON, polishes the prompt when you press Enter, asks Apply Polish, Keep Original, or Retry Polish in Claude Code's question dialog, and sends the one you choose. It has guides for Opus 5.5, Sonnet 5.5, GPT-6 Astra, and GPT-5.6 and picks one from the session's model, and the polish call gets the project's file list, CLAUDE.md, manifests, files the prompt names, and the agent's last answer, never .env, key, or secret files. In the source, the polish is one $.model.complete call, Claude Code's own client.
Demo

What it hooks into
session.startWhen a session startscommand.runWhen you run its slash command/polish/p_polishprompt.submitWhen you send a promptskill.promptA Claude Code eventsession.measureWhen usage is measuredturn.completeWhen a turn endsui.closeWhen its pane closesui.faultA Claude Code eventui.messageAdds a message to the sessionui.renderDraws into the sessionPaneAbovePromptSessionMode
Read from the on() calls in its hook modules. Open its source under Files to check what each one does.
Files
The author's README and source, 4 files. They load when you open them.
Facts
- Author
- jheroorehj
- Category
- Review
- Marketplace
- prompt-polish
- Version
- 0.10.1
- License
- MIT
- Repo stars
- 7
- Last change
- 2026-10-25
- Language
- TypeScript
- Files
- 70