# alchaincyf/huashu-design

Source: [GitHub](https://github.com/alchaincyf/huashu-design)  
Feed7 permalink: https://feed7.dev/p/huashu-design-1tyg5ud  
Published: Unknown  
Trust: Needs Review (needs_review)

## Why Included

Huashu Design packages a multi-agent-friendly design workflow as an HTML-native skill: generate six directions, develop three, then check deliverables and design history. It supports editable PPTX and media exports, with limits around complex animation and blank-brand work.

## Source Summary

Version **3.0** offers an 84-style library with samples and first creates six thumbnails, then develops three directions for review. It can produce clickable app prototypes, HTML decks, editable-text PPTX, infographics, and animation exports. Local preference notes and a visual ledger stay on the user's machine.
Builders who want design work inside coding-agent sessions can use its staged review flow to compare directions before polishing one. The factual lint checks issues such as broken images, unreadable text, low contrast, and console errors; the ledger helps flag repeated visual patterns across projects.
The workflow is HTML-native and does not support complex 3D or physics-based animation. Blank-brand work is described as lower quality, and the PPTX export preserves editable text boxes but does not make designs Figma-layer editable.

## Practical Implication

No separate implication recorded.

## Agent-Ready Context

Version **3.0** offers an 84-style library with samples and first creates six thumbnails, then develops three directions for review. It can produce clickable app prototypes, HTML decks, editable-text PPTX, infographics, and animation exports. Local preference notes and a visual ledger stay on the user's machine.
Builders who want design work inside coding-agent sessions can use its staged review flow to compare directions before polishing one. The factual lint checks issues such as broken images, unreadable text, low contrast, and console errors; the ledger helps flag repeated visual patterns across projects.
The workflow is HTML-native and does not support complex 3D or physics-based animation. Blank-brand work is described as lower quality, and the PPTX export preserves editable text boxes but does not make designs Figma-layer editable.

## Connected Context

Feed7 judgment across 910 accumulated Signals:

Huashu turns agent-assisted design into a staged review workflow: compare six thumbnails, develop three directions, then refine and lint outputs. Its local preference notes and visual ledger support continuity across projects, while HTML-native production, weaker blank-brand results, and limited PPTX editability bound where the workflow fits.

- [Leonxlnx/taste-skill](https://feed7.dev/p/taste-skill-15nf4kv) — Both encode design judgment for coding agents; Huashu adds a staged visual comparison and linting workflow, while taste-skill focuses on inferred design language and adjustable style choices.
- [How our agents build on-brand pages with design.md](https://feed7.dev/p/how-our-agents-build-on-brand-pages-with-design-md-1gt2ft4) — Vercel separates brand guidance, CSS primitives, and repeatable evaluation; Huashu adds persistent local preferences and factual visual lint, giving concrete complementary mechanisms for consistency and review.
- [cathrynlavery/diagram-design](https://feed7.dev/p/diagram-design-1456lje) — Diagram Design applies reusable guidance to accessible HTML and SVG diagrams, whereas Huashu covers broader HTML-native design outputs; their constraints help identify when a focused diagram workflow is more appropriate.
- [The Missing Layer: Design Taste in AI Agents — Hassan El Mghari, Together AI](https://feed7.dev/p/the-missing-layer-design-taste-in-ai-agents-hassan-el-mghari-together-ai-00err2g) — The talk recommends references, focused prompts, and iteration to improve agent-made UI; Huashu operationalizes part of that approach with multiple candidate directions and a review stage, without establishing that its checks settle subjective quality.

## Context Map

- Layer: agent
- Domains: image, video
- Topics: skills, generative-media, design-engineering

## Uncertainty

- None recorded.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.