brand intelligence api · v0.1 · 2026

Hi, we're brandnana.

One CLI call returns a brand book — a single-file workbook with live brand assets, competitor ads, and a deep product catalog. It installs straight into your agent as a Claude Code skill.

open portal → read the docs

works with Claude Code Codex Cursor any MCP client

~/work · brandnana

$ brandnana book nike.com [brandnana] fetching nike.com… ✓ brand identity 1.2s ✓ 3 competitors 2.1s ✓ 247 products 4.6s ✓ 142 ads (meta+tt) 3.4s ✓ design palette 0.8s [brandnana] installed brand book: nike open  ./nike.workbook.html query brandnana book query nike

247 brand books built

89 brands captured

123 MCP tools

0.42ms avg query

~3MB workbook size

§ 01 · what you get

A single workbook your agent can read.

Drops at ~/.claude/skills/brands/<slug>/ with a SKILL.md so Claude knows when to use it. Everything else lives inside one offline-capable .workbook.html.

~/.claude/skills/brands/nike/
├── SKILL.md → tells Claude when to use this skill ├── nike.workbook.html → single-file, gzip-packed, ~3 MB │   ├── brand.org · identity, palette, fonts, voice │   ├── competitors/ · adidas, puma, asics — ad breakdowns │   ├── catalog/ · products by category / color / price │   ├── ads/ · 142 ad captures (streamed or embedded) │   └── timeline.org · what was captured, when, by who └── query.md → OQL examples, scoped to this brand

format: single-file HTML · WASM runtime baked in
data: org-mode tree · queryable via OQL
media: embedded (offline) or R2-linked (slim)
install: drops into ~/.claude/skills/brands/
query: brandnana book query nike '(under "Catalog/by-color/Black")'

§ 02 · six verbs

Six verbs. One CLI. 123 MCP tools.

Each verb returns clean JSON if you call it raw — or feeds straight into the brand book when you call brandnana book.

brand fetch

Pull live brand identity — logos, palette, font stack, voice, social handles.

$ brandnana brand fetch nike.com
✓ primary: #111  ·  secondary: #f5f5f5
✓ fonts: Futura, Inter, Arial
✓ logo: nike-swoosh.svg (24 KB)
✓ tone: bold, declarative, athletic

ads search

Scrape live creative from any brand's ad library — Meta, TikTok, Google.

$ brandnana ads search --domain=adidas.com
› meta:    142 ads  (28 active)
› tiktok:   38 ads  (6 promoted)
› google:   54 ads  (18 with video)
› themes:  return-to-running · trail · y2k

catalog crawl

Extract a structured product catalog from any storefront — Shopify, custom, anywhere.

$ brandnana catalog crawl glossier.com
▸ products: 247 (32 categories)
▸ images:   1,184 (avg 3.6mb)
▸ price band: $14 — $98
▸ variants: 812 · colors: 42

brief generate

Composable creative brief — pulls in brand context + a prompt, drops a tight one-pager.

$ brandnana brief generate "summer launch"
✓ strategy:  position vs Nike Air Max
✓ tone:       playful, surprising
✓ assets:    use SS25 palette
✓ word count: 412 · reading: 1m48s

design palette

Extract a design system palette from any logo, image, or whole brand.

$ brandnana design palette logo.png
● #f0c040  banana       (primary)
● #d65a3a  cherry       (secondary)
● #1f1a14  ink          (text)
● #fbf4e4  parchment    (background)

mcp ready

Drop into Claude Code, Codex, Cursor as a native MCP server. 123 tools, zero config.

$ brandnana mcp
[brandnana] mcp ready on stdio
  ➜ 123 tools exposed
  ➜ skill: ~/.claude/skills/brandnana/
  ➜ try:  use brandnana to research nike.com

§ 03 · drop in

Drop into your agent in 30 seconds.

brandnana ships an MCP server too. One config block — and your agent has 123 tools to fetch brands, search ads, crawl catalogs, generate briefs.

// ~/.claude/mcp.json
{
  "mcpServers": {
    "brandnana": {
      "command": "brandnana",
      "args": ["mcp"]
    }
  }
}

Your first brand book in under a minute.

No accounts. No SDKs. Just a CLI and an agent that already knows how to use it.

curl -fsSL https://install.brandnana.net | sh

Hi, we're brandnana. 🍌