Claude Skill

Voice Cloner

Feed it tweets or transcripts, and it builds a reusable voice profile your marketing posts can write from.

The Voice Cloner is a Claude skill that reads a person's or brand's existing writing and captures how they write into one reusable Voice Profile. Hand it a folder of tweets, a transcript, a stack of LinkedIn posts. Get back a style guide your other writer skills can read and write from.

The Problem

You've got a voice in your head. Maybe it's yours, maybe it's a founder you ghostwrite for, maybe it's a brand you're trying to sound like. You know it when you read it. The short sentences. The way they open cold with a number. The fact that they never, ever use an emoji.

So you sit down to write a post in that voice. Three lines in, it sounds like you. Generic. Safe. You go back, pull up twelve of their old tweets, read them again, try to absorb the rhythm by osmosis. You write another draft. Closer, but the opener is wrong and the closer reads like a LinkedIn motivational quote they'd never post. You paste two more examples into the chat, re-explain "make it punchier, drop the hashtags," and hope it sticks for the next one. It doesn't. Next session, you start over from scratch.

Every writer skill you use is flying blind. None of them remember the voice. You're the memory, and you're re-teaching it every single time.

This skill ends that. Study the writing once, and the voice becomes a file every other skill can read.

What this Claude skill does

  • Reads real samples in any format: Markdown, .txt, .csv, .xlsx, .json, or a whole folder of mixed sources. Tweets, comments, LinkedIn posts, YouTube transcripts, newsletters.
  • Extracts the actual signal: tone, vocabulary signature, sentence rhythm, hook patterns, closer patterns, recurring themes, formatting habits, and the hard "never do this" rules.
  • Writes a structured Voice Profile Built for two readers: a human skimming it, and another skill pulling from it.
  • Drops in a voice-prompt block that twitter-x-post-writer, linkedin-post-writer, and the video script generator read verbatim to write in that voice without re-analyzing anything.
  • Refines in place when you re-run it: one canonical file per voice, sharpened with each new batch of samples. Never a -v2.
  • Quotes only what it can find: every example line traces back to a real source file. No invented quotes, no smoothing.

Why it's amazing!

  • Teach a voice once, then every writer skill speaks it.
  • Stop pasting the same five example tweets into every new chat.
  • The profile gets sharper over time instead of resetting each session.
  • It captures the voice as it really is, typos and loose punctuation included, not a cleaned-up version.

Who it's for

Anyone who writes in someone else's voice for a living. Ghostwriters, content marketers, social media managers, founders building a personal brand, and agencies juggling a dozen client voices at once. Anyone who's ever re-explained "make it sound more like them" to a blank chat for the hundredth time.

How it works

Point Claude at the writing:

  • "Clone the voice in alex-hormozi-x-posts.md."
  • "Build a voice profile from this folder of my LinkedIn posts."
  • "Study how this person writes and capture it: [paste transcript]."
  • "Analyze my last 300 tweets and write a style guide I can reuse."
  • "I ghostwrite for a SaaS founder. Here's a CSV of his comments. Extract his voice."
  • "Add these new newsletters to his existing voice profile and sharpen it."

You get back a profile file with sections like:

  • Voice in one paragraph — enough to draft in the voice off a single read.
  • Reusable Prompt Block — a compact voice-prompt brief other skills inject directly.
  • Tone & sentiment — register, warmth, confidence, edge, humor, scored and justified.
  • Vocabulary signature — the words they reach for, the words they never use, casing and profanity habits.
  • Hook & closer patterns — named, each with a verbatim opener or closer pulled from the source.
  • Anti-patterns — the hard "never" rules that would read as fake.
  • Examples — 6 to 12 verbatim lines, each tagged to its source file.
  • Confidence & gaps — sample size, what's well-evidenced, what needs more data.

Re-run it later with new samples and it updates the same file in place, logs what changed in a changelog, and bumps the source manifest. The voice gets more accurate every pass.

Pairs well with these other skills

A Voice Profile is only worth building because of what reads it. These skills sit on both ends of it — the writers that pull from the profile, and the scrapers that gather the samples to build it:

FAQ

Does it cost anything to run?

Only a Claude subscription. Other than that it reads files you already have and writes a markdown profile. No external APIs to run this skill.

Will it make up quotes to fit a pattern?

No. Every example line and every verbatim opener, closer, or theme quote has to be findable in a source file. If it can't find a clean line for a pattern it noticed, it describes the pattern in words and leaves the quote out.

Does it research the person, like their bio or background?

No. It only analyzes the writing samples you give it. For biographical research on a real person, use /people-osint-researcher instead.

Does it actually write posts in the voice?

No. It captures the voice into a profile file. The writing happens in twitter-x-post-writer, linkedin-post-writer, or the video script generator, which read the profile and write from it.

How is this different from an ICP?

An ICP profiles your customer. This profiles a writer's style. If you want to define who you're writing for, use /icp-generator. If you want to capture how someone writes, use this.

How many samples does it need?

More is better, and it tells you. A profile built from 300 tweets is high-confidence on hooks and formatting but may be thin on long-form rhythm. Add a transcript or two and the confidence note updates.

What if I gather more samples later?

Re-run it on the same voice. It reads the existing profile, folds in the new samples, updates every section, and logs the change. One file per voice, sharper each time.

What file formats can it read?

Markdown, plain text, CSV, XLSX, JSON, and transcripts in any of those. Point it at a single file or a whole folder and it ingests everything supported.

Safe To Install

Please read before installing. Any skill you copy online is someone else's code running inside your Claude, and a lot of what's out there should give you pause. Most authors who give skills away don't give a second thought to quality or security. Most are vibe-coded slop pushed out untested, and some are the occasional bad actor hiding an exploit in plain sight. (Read about OpenClaws ClawHub security nightmare). Every skill we publish is vetted by a programmer with 20+ years of professional experience and checked line by line for anything that touches your files without reason. With Stim-Pack Studios Claude Skills, you get peace of mind knowing you're installing safe, vetted code, not a stranger's guess.

Add it to your Claude workflow

If you write in other people's voices, this is the skill that makes the rest of your writer skills worth using. It turns "sound more like them" from a thing you re-explain every session into a file every skill can read.

Get the Voice Cloner Claude skill and capture a voice once instead of teaching it forever.