xx-voice-apply

name	xx-voice-apply
description	apply a voice style

Voice Apply

Rewrites a document to match a specific writing voice, guided by a style guide.

Arguments

• <slug> — voice identifier (e.g., xian). Resolves to ~/.personal/<slug>-writing-style.md.
• <document-path> — path to the document to rewrite.

Flow

1. Resolve voice file: Look for ~/.personal/<slug>-writing-style.md.

   • If ~/.personal/ doesn't exist: tell the user to create it: mkdir -p ~/.personal
   • If the style file doesn't exist: tell the user to generate one first: /xx-voice-gen <slug> <folder-of-sample-writing>
   • Do NOT proceed without a valid style file.

2. Read the style guide — load the full voice file. This contains specific patterns, examples, vocabulary, sentence structure, and anti-patterns.

3. Read the document — load the target document.

4. Rewrite in two passes:

   • Pass 1: Content audit. Read the document for structure and argument. Do not change anything yet. Note which sections feel off-voice.
   • Pass 2: Voice rewrite. Rewrite the document applying the style guide. Preserve the content, structure, and meaning. Change the voice: sentence structure, word choices, openings, closings, transitions, emphasis patterns. Use specific patterns from the style guide, not vague "make it more conversational."

5. Show the diff — present the changes so the user can review before accepting.

Rules

• Preserve content. The rewrite changes voice, not substance. Don't add or remove arguments, examples, or sections unless the style guide specifically calls for it (e.g., "never write long concluding summaries").
• Be specific. When applying a style pattern, you should be able to point to a rule in the style guide that justifies the change.
• Don't over-apply. Not every sentence needs every pattern. The style guide describes tendencies, not rigid rules. A document that hits every pattern in every paragraph will feel forced.
• Respect the document type. A letter to a CEO and a blog post have different constraints even in the same voice. Adapt the intensity of voice application to the context.