whystrohm-voice-extract
Extract a 6-dimension voice profile from any URL. Generate 15-20 enforceable guardrails. Outputs as CLAUDE.md.
Style guides are PDFs nobody reads. Brand voice should be code. voice-extract pulls voice from any URL and outputs it as enforceable rules.
The Problem
Every brand has a 40-page style guide. None of them are enforceable. The PDF lives in Notion, the AI tools generate slop, nobody connects the two.
What It Does
- 01Crawls any URL and extracts a 6-dimension voice profile
- 02Generates 15-20 starter guardrails ready to enforce
- 03Outputs as CLAUDE.md any LLM can load
The 6 dimensions
voice-extract scores brand voice across six axes, each calibrated 0-100:
- Authority — assertive vs hedging
- Emotional temperature — warm vs cold
- Proof density — specific numbers vs general claims
- Cadence — short punchy vs long flowing
- Vocabulary range — accessible vs technical
- Rhetorical structure — narrative vs argumentative
These six numbers become the rubric every downstream skill enforces.
What gets generated
The output lands as a CLAUDE.md you can drop into any project root. It contains:
- The 6 axis scores with one-line interpretations
- 15-20 starter guardrails (forbidden words, required structures, banned openings)
- 5-10 exemplar sentences pulled from the source
- A "voice in one paragraph" summary for human review
Any LLM that loads the file will read these as constraints on every subsequent generation.
Install
git clone https://github.com/whystrohm/whystrohm-voice-extract.git ~/.claude/skills/whystrohm-voice-extract
Full docs on GitHub →
How It Composes
First step in the voice pipeline. Outputs the brand-config.json that whystrohm-audit, voice-scorer, shotkit, and slopfiles all read from. One extraction; consistent voice across every downstream skill.
Related Skills
media-tsunami
The empirical layer. Extracts brand voice as executable code — cadence, vocabulary, forbidden words, exemplar sentences — serialized as a CLAUDE.md any LLM can load.
Install →whystrohm-audit
Score any content against a 5-layer infrastructure framework. Diagnose what's broken. Rewrite one piece live.
Install →whystrohm-voice-scorer
Measure voice drift between your website and social content. Find where the brand is leaking.
Install →