| name | brand-visual-asset-system |
| description | Reusable brand visual asset system for premium content deliverables. Use when creating or redesigning PPT, HTML, PDF, Word, poster, brand deck, product introduction, commercial proposal,招商/路演/品牌方案, or marketing content files AND the user asks for high-end visual design, brand-handbook quality, product/brand storytelling, GPT Image 2 generated imagery, or references to the private excellent PDF visual assets. Do not use for ordinary document conversion, plain Word/PDF text drafting, contracts, meeting notes, technical docs, tables, or file tasks where no brand/visual/premium design intent is present. |
Brand Visual Asset System
Use this skill as a passive Office / Creative Route skill. It converts a content brief into a visually polished brand-style deliverable by reusing distilled visual principles from three private reference PDFs, not by copying their original assets.
Trigger Boundary
Use only when both conditions are true:
- The deliverable is a content file or visual communication artifact: PPT, HTML, PDF, Word, poster, brand deck, product introduction, proposal, campaign page,招商材料, or similar.
- The user signals visual or brand intent: high-end design, brand feel, polished layout, excellent PDF reference, product storytelling, GPT Image 2 imagery, brand-handbook style, commercial proposal, marketing, retail, FMCG, lifestyle, or similar.
Do not use for plain file generation, simple format conversion, ordinary reports, legal/contract drafting, meeting notes, technical documentation, spreadsheets, or any task where the user only needs content and not visual direction.
Copyright Boundary
Treat the three reference PDFs as private visual references only.
Reuse:
- Visual DNA
- Layout logic
- Color relationships
- Page rhythm
- Content module patterns
- Image prompt structures
Do not reuse:
- Original PDF images
- Logos or proprietary brand marks
- Celebrity or real-person portraits
- Media screenshots
- Exact commercial data
- Any third-party proprietary visual asset
Generate new imagery with gpt-image-2 when imagery is needed, then typeset final text in the target format.
Workflow
- Classify the target output: PPT, HTML, PDF, Word, poster, brand deck, or product/commercial proposal.
- Read
references/reuse-pipeline.md for the production flow.
- Choose one primary visual DNA from
references/visual-dna.md.
- Choose page templates from
references/layout-recipes.md.
- If images are needed, read
references/gpt-image-2-briefs.md and use only gpt-image-2 prompts or the gpt-image-2-content-factory skill.
- Assemble the deliverable using the relevant file/document/presentation/PDF tooling.
- Check that ordinary source-PDF assets were not copied and that the final page has clear hierarchy, no overlaps, and format-appropriate readability.
Visual DNA Choices
retro-street-retail: for fashion, sports, shoes, retail, national-trend, store-space, product matrix, and channel display.minimal-symbol-narrative: for brand attitude, user insight, lifestyle brands, concept storytelling, symbol systems, and clean chapter rhythm.youth-fmcg-proof: for beauty, nutrition, food, health, FMCG, product proof, sales data, user profile, KOL/media proof, and young consumer brands.
Use one primary DNA and at most one supporting DNA. Do not average all three styles in one deliverable.
Format Guidance
- PPT: Use strong visual pages and proof/information pages in alternation.
- HTML: Use full-screen hero/sections, CSS grid, and responsive stacking.
- PDF: Use HTML/CSS rendering when design fidelity matters; check fonts, margins, blank pages, and image clarity.
- Word: Preserve editability; use cover, headings, image captions, tables, and restrained brand visual pages rather than forcing full-slide layouts.
For PAISHU-branded deliverables, also apply paishu-content-file-design unless the user gives a newer explicit style instruction.
Final Checks
- The skill triggered for a brand/visual/premium-content reason, not merely because a file type was mentioned.
- The selected DNA and page recipes match the content's business purpose.
- Any image prompt uses only
gpt-image-2.
- Text-heavy parts are typeset outside the generated image whenever exact wording matters.
- The result does not copy original PDF visual assets.
- The target format remains usable: editable PPT/Word when requested, renderable HTML/PDF when requested.
