> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mossx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompt enhancer

> Rewrite the input with a chosen provider and model before you send, so the instruction is more specific.

Prompt enhancer rewrites the current input before you send. The original sentence can be short. After the rewrite, the model you are chatting with runs it.

## When to use it

It helps with drafts like:

* "Make the login look a bit nicer"
* "Performance feels off"
* "Take a look at this"

Skip it when the instruction already has acceptance criteria and a file scope. An extra pass only spends tokens.

## How to use it

1. Write a draft in the input box. You can include `@` file mentions
2. Click the enhance control on the input toolbar
3. Review the rewrite and edit it if needed
4. Send

The enhancer provider and model do not have to match the current chat. For example, chat with Claude Opus and enhance with a faster Haiku.

## Configure it

Open **Settings → Prompt enhancer**.

You can set:

* Which provider to use (Claude Code or Codex)
* Which model to use
* Automatic choice or a manual pick
* Restore defaults

If the current provider is unavailable, the page marks it. Configure that runtime in [Provider management](/en/vscode/providers/overview) first.

General also shows a prompt-enhancer summary. The full options are still on this page.

<Note>
  Enhancing itself uses tokens. Use it when the draft is vague enough to be worth cleaning up first.
</Note>