Skip to content

SlashImagine/tonethief

Repository files navigation

🎙️ tonethief

Steal any brand's voice. Instantly.

Point it at any URL. Get a deployable VOICE.md in seconds — archetypes, tone spectrum, vocabulary DNA, and a copy-paste AI prompt that makes any LLM write indistinguishably in that brand's voice.

Zero API keys. Zero dependencies. Zero cost.

🌐 Website


Quick Start

npx tonethief https://liquiddeath.com

That's it. You get a complete VOICE.md — drop it into any AI chat, agent, or system prompt.

The Output

Default — A deployable VOICE.md

Not a report. An operating manual. Drop it into any project and your AI writes as that brand.

tonethief https://liquiddeath.com --output VOICE.md

Includes:

  • Identity — archetypes, traits, one-liner
  • Tone spectrum — "extremely irreverent (9.9/10)", "notably playful (7.6/10)"
  • Writing rules — sentence structure, vocabulary targets, reading level
  • Do/Don't guidelines — specific to the brand, not generic advice
  • Brand-specific patterns — dark humor, profanity, death themes, ALL CAPS as style
  • Voice test — 4-question checklist before publishing
  • System prompt — copy-paste into any LLM

--report — Full analytical breakdown

tonethief https://stripe.com --report

Compare — Side-by-side brand diff

tonethief compare https://stripe.com https://liquiddeath.com

JSON — For pipelines

tonethief https://notion.so --format json --output notion.json

10-Dimension Tone Spectrum

Dimension What it measures
Formal ↔ Casual Contractions, slang, register
Serious ↔ Playful Humor, wit, entertainment value
Technical ↔ Accessible Jargon density, complexity
Reserved ↔ Enthusiastic Energy, superlatives, exclamations
Corporate ↔ Human "Synergy" vs "you and me"
Passive ↔ Active Voice construction
Vague ↔ Specific Numbers, data, concrete claims
Long-winded ↔ Concise Sentence length, economy
Conventional ↔ Irreverent Profanity, subversion, punk energy
Safe ↔ Provocative Dark humor, shock value, edge

12 Personality Archetypes

The Friend · The Expert · The Cheerleader · The Storyteller · The Analyst · The Coach · The Minimalist · The Doer · The Rebel · The Jester · The Provocateur · The Maverick · The Sage

Smart Crawling

tonethief doesn't just scrape the homepage:

  • Prioritizes voice-rich pages — /about, /story, /mission, /manifesto, /blog
  • Semantic zone weighting — hero copy 3x, headings 2x, nav/footer 0x
  • Content deduplication — catches locale variants with identical content
  • Auto-seeds common voice paths when link discovery fails
  • Default 8 pages (configurable up to 20)

Options

Flag Description
--report, -r Full analytical markdown report
--format, -f Output format: voice (default) | markdown | json
--pages, -p Max pages to crawl (default: 8, max: 20)
--output, -o Write to file instead of stdout
--verbose, -v Show crawling progress
--help, -h Show help
--version Show version

Programmatic API

import { analyzeBrandVoice, formatVoiceFile, formatMarkdown, formatJSON } from 'tonethief';

const profile = await analyzeBrandVoice('https://liquiddeath.com', { maxPages: 8 });

// Get the VOICE.md (default)
const voiceMd = formatVoiceFile(profile);

// Or the full report
const report = formatMarkdown(profile);

// Or raw data
const json = formatJSON(profile);

How It Works

  1. Crawl — Fetches up to 20 pages, prioritizing voice-rich content (about, blog, manifesto)
  2. Extract — Semantic zone weighting amplifies hero copy and headings, suppresses nav/footer
  3. Analyze — 10-dimension tone analysis using linguistic heuristics (word patterns, sentence structure, vocabulary fingerprinting)
  4. Classify — Maps to personality archetypes and derives brand-specific patterns
  5. Generate — Produces a deployable VOICE.md with writing rules, guidelines, and AI prompts

No AI APIs called. No data sent anywhere. Everything runs locally.

License

MIT


Built by Ad Machine · Star on GitHub

About

Steal any brand's voice from their website. Deployable VOICE.md in seconds. Zero deps, zero API keys.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors