Point it at any URL. Get a deployable VOICE.md in seconds —
archetypes, tone spectrum, vocabulary DNA, and a copy-paste AI prompt.
How it works
Give it any URL. Homepage, about page, blog. It finds the pages that matter.
Crawls 8 pages. Runs 10-dimension tone analysis. Detects archetypes, vocabulary, rhythm.
You get a VOICE.md — drop it into any project or AI agent and it writes in that voice.
The output
Most tools give you analytics. tonethief gives you something you can actually use — a deployable brand identity file.
Drop VOICE.md into any project, AI agent, or system prompt. Your AI will write indistinguishably from that brand.
Think SOUL.md — but for any company on the internet.
What you get
12 types mapped — Rebel, Outlaw, Jester, Sage, Minimalist, and more. The ones that fit, ranked.
Formal/Casual, Serious/Playful, Earnest/Irreverent, Dry/Humorous. Each scored 1–10.
Power words, reading level, sentence rhythm. The fingerprint of how they actually write.
Concrete writing rules extracted from real brand copy. Not generic advice — theirs.
Drop into ChatGPT, Claude, or any agent. Instantly write in any brand's voice.
Side-by-side voice diff. Run tonethief compare stripe.com square.com and see exactly where they diverge.
Get started
$ npx tonethief https://yourbrand.com
No install required. Runs immediately. Zero API keys.
$ npm install -g tonethief
$ tonethief https://yourbrand.com --output VOICE.md
$ git clone https://github.com/SlashImagine/tonethief
$ node bin/cli.js https://yourbrand.com