business-branding - SKILL.md Agent Skill

name: business-branding description: Use this skill when a business owner wants marketing, branding, or advertising assets — logos, color palettes, typography mood, social posts, display ads, or short promo videos. Runs a hybrid intake (structured questions for vibe, style, color scheme, and logo type; free-form for audience, competitors, and inspirations), translates the answers into a reusable brand system, then orchestrates the pixio-skill to generate the actual assets. Trigger for "build my brand", "design a logo for my business", "create ads for my company", "make social media posts for my shop", "brand identity for my startup", and similar phrasings. Do not use for one-off generic image prompts with no business context — use pixio-skill directly for that.

Business Branding Skill

This skill turns a business owner's intent into a coherent set of brand and marketing assets. It does NOT call the Pixio API itself — it runs intake, builds a reusable brand system, then invokes pixio-skill for each generation.

Default Deliverables

Produced by default (in this order):

Brand identity — 3 logo concepts, a 5-swatch color palette, and a typography mood reference.
Social media posts — 3 starter posts in 1:1 (feed) and 9:16 (story) formats.
Ad creatives — 1 hero banner (16:9) and 1 square display ad (1:1).

Produced only when the user explicitly asks:

Short video ads — 5–10s text-to-video or image-to-video promos. Video burns significantly more credits, so confirm scope and budget before starting.

Workflow

Intake. Load references/intake.md and run it. Use AskUserQuestion for the 4 structured questions (vibe, style, colors, logo type). Then ask the free-form questions conversationally.
Build the brand system. Load references/brand-system.md and translate intake answers into a reusable prompt prefix, palette description, and style descriptor. Echo the brand system back to the user as a short summary and get sign-off before generating anything.
Confirm API readiness. Ask for a Pixio API key if not already in context. Mention the default Pixio account concurrency is 1 in-flight generation (3 for Maker accounts) — so deliverables will be produced sequentially, not in parallel.
Generate brand identity first. Load references/guides/logo.md. Generate logo concepts before anything else — the chosen logo and refined palette feed every later asset.
Get sign-off on the logo + palette. Show the user the outputs, let them pick a favorite or request iterations.
Generate the rest of the deliverables. Use references/guides/social-posts.md and references/guides/ad-creatives.md. Reuse the brand system prompt prefix in every generation for visual consistency.
Optional: video. Only if the user asked, load references/guides/video-ads.md.
Hand off. Return all output URLs grouped by deliverable, plus the brand system summary as a reusable spec the user can save.

How To Invoke pixio-skill

When it's time to actually generate, invoke the pixio-skill via the Skill tool. Pass it a self-contained generation brief — model intent (text-to-image / image-edit / text-to-video), aspect ratio, the full prompt (brand system prefix + asset-specific prompt), and any reference image URLs from prior generations. The pixio-skill handles model discovery, params, uploads, and polling.

Do NOT try to call Pixio endpoints directly from this skill. If you find yourself writing curl https://beta.pixio.myapps.ai/..., stop and delegate.

Reference Routing

references/index.md — start here when unsure which doc to load.
references/intake.md — the hybrid questionnaire.
references/brand-system.md — translating answers into a reusable prompt prefix.
references/guides/logo.md — logo concepts (3 variants per type).
references/guides/social-posts.md — feed + story templates, multi-platform sizing.
references/guides/ad-creatives.md — banner, square, hero ad formats.
references/guides/video-ads.md — opt-in video promos.
references/examples/end-to-end.md — one worked example from intake through first deliverable.

Guardrails

Never start generating before intake + brand-system sign-off. Skipping intake produces incoherent assets and burns credits.
Reuse the brand system prefix in every prompt. This is what makes the assets feel like one brand instead of ten random images.
Logo first, always. The logo informs palette refinement and downstream prompts.
Confirm before video. Video is opt-in and costs 5–20x a single image.
Don't invent trademarks. If the user names their business after an existing brand, flag it before generating identity assets.
Don't fabricate model IDs or params. This is the pixio-skill's job — let it do model discovery.

Completion Checklist

Before declaring the brand package ready:

Intake answers (all 4 structured + free-form) are captured and the brand system was signed off.
Logo concepts were generated and one was selected (or the user explicitly asked to keep all 3).
Color palette and typography mood are documented as text the user can paste into a brief.
Social posts and ad creatives were generated with the brand system prefix.
Video was generated only if explicitly requested.
All output URLs are grouped by deliverable in the final response.
Brand system summary is included so the user can re-run later with consistency.