doubao-creative-design

Generates and edits commercial creative images via routed text-to-image and image-to-image workflows.

2|Updated Aug 9, 2026
One-click install
npx skills add https://github.com/DeepJH/doubao-skill-and-info --skill doubao-creative-design-deepjh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doubao-creative-design
Source: https://github.com/DeepJH/doubao-skill-and-info/tree/main/skills/doubao-creative-design
Command: npx skills add https://github.com/DeepJH/doubao-skill-and-info --skill doubao-creative-design-deepjh

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It turns vague image requests into structured, production-ready visual assets by routing each request to the right design ruleset and enforcing consistency across multi-image deliverables. ## Core Features & Use Cases - Intent Routing: Classifies requests into Logo systems, IP characters, packaging, brand applications, e-commerce visuals, knowledge/education graphics, or social media creatives, loading only the relevant reference rules. - Core Asset Consistency: Locks a core asset (Logo, IP, product, key visual) and derives all subsequent images from it via image editing, preserving identity across series and multi-ratio adaptations. - Prompt Governance: Enforces rules for renderable text, ratios, clarification limits, image count strategy, and post-generation review. - Use Case: A user asks for a brand mascot plus a poster series and social media covers; the Skill generates the core IP first, then derives consistent turnarounds, posters, and platform-specific adaptations. ## Quick Start Ask the assistant to design a brand mascot with three-view turnaround and a matching promotional poster series.

Frequently Asked Questions about doubao-creative-design

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a consistent brand IP character across multiple images?

First generate or confirm a core IP image, then derive all subsequent assets (turnarounds, posters, stickers, merchandise) from that core image URL using image editing. This preserves face structure, proportions, colors, and signature props across the whole set.

How does the skill choose between text-to-image and image-to-image?

Text-to-image is used only to create a first core asset from scratch. Any derivative work based on uploaded images, prior generations, or locked core assets must use image editing with the reference image URLs passed in.

What types of design deliverables does this skill support?

It covers Logo systems, IP character systems, packaging, brand application materials, e-commerce main images and detail pages, knowledge and education visuals, and multi-platform social media creatives, each with dedicated reference rules.

Can it generate Amazon-compliant product images?

Yes. Amazon main images follow strict pure-white-background rules with no extra text or badges, while infographic listing images and A+ modules carry objective, detailed copy. Prices and promotional claims are prohibited in Amazon assets.

When does the skill ask clarifying questions before generating?

It asks at most one round of one to three questions, only when missing information would change the fundamental direction, such as an undefined subject, unclear product category, or missing required values like dates or prices.

What are the limitations on text rendered inside generated images?

All visible text must be explicitly written out and wrapped in Chinese double quotes; placeholders like XX are forbidden. Prices, discounts, dates, certifications, and rankings may only come from user-provided content and are never fabricated.