brand-pack-reader

name	brand-pack-reader
description	Use this skill at the start of any Shopify store build session, ad creative session, or design task that needs brand-consistent output. It defines exactly how Claude ingests a brand pack from brands/<brand>/brand-pack.md and applies it to copy, colors, typography, pricing, and section content. Activates whenever a brand-pack.md is in context or a new build session begins.

Brand Pack Reader

Purpose: Define exactly how Claude ingests a brand pack and applies it across copy, colors, and structure decisions — ensuring zero hallucination of brand details.

---

WHAT A BRAND PACK CONTAINS (standard structure)

When Claude reads brands/<brand>/brand-pack.md, it extracts and maps:

Brand Pack Section	What Claude extracts	Where Claude uses it
Brand name + tagline	Display name, CTA phrasing	All copy, button labels, meta
Color palette (hex codes)	Exact hex values + token names	settings_data.json, section backgrounds
Typography	Font names + weights	settings_data.json, layout decisions
Pricing	Price + compare-at	Product settings, ATC button labels
Announcement bar text	Exact copy + colors	Announcement bar section
Emoji benefit bullets	5 pills with emoji + label	main-product emoji_benefits block
Sticky ATC	Button label + sub-label	sticky_atc block
Buy buttons	Primary label + save callout	buy_buttons block
Shipping line	Shipping text	estimated_delivery block
Trust icons	Three icons + labels	Below ATC trust bar
Founder/About copy	Source for About block	About section
Headlines and proof points	Copy for hero, transformation, testimonials	Across PDP
Risk reversal / guarantee	Exact phrasing	Guarantee section + below CTA
FAQ Q&A	Question + answer pairs	Collapsible FAQ section

---

STRICT RULES

1. Never invent brand details. If the brand pack doesn't specify, mark [VERIFY] and ask Jake.
2. Hex codes are exact. Don't approximate "navy blue" → use the exact hex from the pack.
3. Token names match brand pack. If the pack calls a color cast_iron_navy, use that name in settings_data.json — don't rename to primary or brand-color.
4. Copy is verbatim, not paraphrased. When the brand pack gives exact wording (announcement bar, CTA labels, headlines), use exactly that. If you want to suggest an alternative, mark it as a suggestion separately.
5. Aesthetic constraints are binding. If the brand pack says "no emoji" or "no clinical imagery," obey.

---

EXTRACTION PATTERN

When loading a brand pack at session start:

Step 1: Identify brand identity block (name, product, USP, core promise)
Step 2: Pull color palette table → store hex codes and token names
Step 3: Pull typography (headings, body, buttons + fallbacks)
Step 4: Pull pricing (price, compare-at, status)
Step 5: Read announcement bar (text + colors)
Step 6: Pull PDP section content (benefit bullets, sticky ATC, buy buttons, etc.)
Step 7: Pull risk reversal + guarantee
Step 8: Pull FAQ Q&A
Step 9: Note any aesthetic rules ("no emoji," "no clinical visuals," etc.)
Step 10: Confirm to Jake: "Brand pack loaded — [brand name], [current campaign]. Ready."

---

OUTPUT VALIDATION

Before delivering any brand-touching output, run this silent check:

• Every color referenced exists in the brand pack
• Every font referenced exists in the brand pack
• Every price/discount matches the brand pack
• No copy invented beyond what the brand pack authorizes
• No aesthetic violations (clinical, generic, off-tone)

If any fail → revise before delivery.

---

CONFLICT HANDLING

If the brand pack contradicts itself (e.g., two different headlines for the same section):

1. Flag the contradiction
2. Show both versions
3. Ask Jake which is current
4. Don't choose silently

If the brand pack contradicts the active campaign offer brief:

1. Flag the contradiction
2. Default to the offer brief (more recent intent) but verify with Jake
3. Suggest updating the brand pack to match