Brand Lock Extractor

Turns a site, PDF, or screenshots into a sourced brand-lock.md with confidence flags on every value.

Testowano · Działa ★ 9.6/10

Co robi ten skill

Extracts a nine-section brand-lock.md (identity, palette, type, mood, never-list, motion, voice) from a website, brand-book PDF, screenshots, or written description, tagging every value extracted/inferred/needs-confirmation with its source. Triggers on 'extract my brand', 'build a brand lock', or handing over brand assets; explicitly declines to run on a vague idea with no assets, pointing the user at the blank template instead.

Raport z testu

Fed it a real site excerpt (CSS custom props + copy); the baseline free-hand pass fabricated a hex (#F5F0E8) and two font names never present anywhere in the source, exactly the failure this skill's Rule 1/2 exist to prevent, while the skill run pulled the real 4 hexes and 2 font families from the CSS and correctly flagged the unstated divider color and font weights as needs-confirmation.

Testowano: 2026-07-15 · Claude Code 2.x (agent harness)

Instalacja

git clone https://github.com/whystrohm/shotkit
mkdir -p ~/.claude/skills
cp -r shotkit/skills/brand-lock-extractor ~/.claude/skills/brand-lock-extractor

Komendy i przykładowe prompty

/brand-lock-extractorTurns a site, PDF, or screenshots into a sourced brand-lock.md with confidence flags on every value.

Skille uruchamiają się na zwykłe polecenia — bez komend do zapamiętania. Po instalacji aktywują go prompty takie jak te (po angielsku):

Build a brand lock from our website, here's the URL
I have a PDF brand book, extract our palette and voice from it
Make a brand pack from my site's colors, fonts, and copy

Pełny raport →