A 5-stage workflow: design URL/image → spec → plan → subagent-driven implementation → playwright TDD verification → visual QA. Self-contained.
Entry point for the design-to-code workflow. Use when the user provides any visual source (design URL, attached image, Figma/Sketch screenshot, an already-rendered page in this project, etc.) AND expresses intent to implement or extend it ("implement this", "build this", "extend this page", or equivalent intent in any language). Dispatches a design-explorer subagent to inspect the visual source, conducts interview-style Q&A, and produces a user-approved spec.md. MUST be followed by design-to-code:writing-plans.
Use after design-to-code:writing-plans completes and plan.md is user-approved. Dispatches fresh implementer subagents per task; each task passes spec-review and code-quality-review before completion. Main agent orchestrates; it never edits code. MUST be followed by design-to-code:tdd-verify-from-spec.
Use after design-to-code:subagent-driven-development completes. Dispatches one ac-verifier subagent per acceptance item in spec.md; failures dispatch fixer subagents. MUST be followed by design-to-code:visual-qa-from-design.
Use after design-to-code:tdd-verify-from-spec completes. Dispatches one view-verifier subagent per view to screenshot and compare against the original design source; dispatches visual-fixer subagents for code-fixable issues. Terminal skill of the design-to-code workflow. Must be used any time the user asks to "验收样式", "对比设计稿", "还原度检查", or after tdd-verify-from-spec finishes.
Use after design-to-code:brainstorming-from-design completes and spec.md is user-approved. Reads spec.md and produces plan.md decomposed into subagent-ready functional units. MUST be followed by design-to-code:subagent-driven-development.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that turns any visual source (design URL, attached image, or an already-rendered page) plus explicit implementation intent into a verified, pixel-faithful implementation — through a 5-stage, linear, spec-driven workflow.
Install from any directory on any machine with Claude Code. You do not need to clone this repository yourself — Claude Code fetches the marketplace from git.
github.com reachablegh auth login or set GITHUB_TOKEN / GH_TOKEN/plugin marketplace add zhangtianwei1998/design-to-code
claude plugin marketplace add zhangtianwei1998/design-to-code
Defaults to user scope (~/.claude/), so it works in any project directory.
/plugin install design-to-code@design-to-code
claude plugin install design-to-code@design-to-code
/reload-plugins
After install, skills are namespaced with the plugin name, for example:
/design-to-code:brainstorming-from-design
List available skills in /plugin → installed design-to-code.
/plugin marketplace update
/plugin update design-to-code@design-to-code
To prompt teammates when they trust a project folder, add to that repo’s .claude/settings.json:
{
"extraKnownMarketplaces": {
"design-to-code": {
"source": {
"source": "github",
"repo": "zhangtianwei1998/design-to-code"
}
}
},
"enabledPlugins": {
"design-to-code@design-to-code": true
}
}
flowchart TD
USER(["User: visual source\n+ implementation intent"])
subgraph S1["① brainstorming-from-design"]
direction LR
a1["Ingest visual source\n(playwright / multimodal)"] --> a2["Interview user"] --> a3["Write spec.md\n+ user approval"]
end
subgraph S2["② writing-plans"]
direction LR
b1["Read spec.md"] --> b2["Explore codebase"] --> b3["Write plan.md\n+ user approval"]
end
subgraph S3["③ subagent-driven-development"]
direction LR
c1["Dispatch implementer\nsubagent per task"] --> c2["Spec review"] --> c3["Code quality review"]
c3 -. "fail → retry" .-> c1
end
subgraph S4["④ tdd-verify-from-spec"]
direction LR
d1["Drive playwright\nagainst running app"] --> d2{"Pass?"}
d2 -- "fail" --> d3["Dispatch fixer\nsubagent"] --> d1
d2 -- "pass" --> d4["✅ Record in\nverify.log.md"]
end
subgraph S5["⑤ visual-qa-from-design"]
direction LR
e1["Load design reference\n(multimodal / playwright)"] --> e2["Screenshot each view"] --> e3{"Visual match?"}
e3 -- "discrepancy" --> e4["Dispatch visual-fixer\nsubagent"] --> e2
e3 -- "match" --> e5["✅ Record in\nvisual-qa.md"]
end
DONE(["Implementation verified\nfunctionally + visually ✅"])
USER --> S1
S1 -->|"spec.md"| S2
S2 -->|"plan.md"| S3
S3 -->|"code + progress.md"| S4
S4 -->|"verify.log.md"| S5
S5 --> DONE
| # | Skill | What it does | Consumes | Produces |
|---|---|---|---|---|
| 1 | brainstorming-from-design | Opens design in playwright or reads images multimodally; interviews you question-by-question; writes a user-approved spec | URL / image / in-project page | spec.md |
| 2 | writing-plans | Explores the codebase, decomposes the spec into subagent-sized tasks with explicit file lists and acceptance criteria | spec.md | plan.md |
| 3 | subagent-driven-development | Dispatches a fresh implementer subagent per task; each task passes spec-review then code-quality-review before moving on | plan.md | code + progress.md |
| 4 | tdd-verify-from-spec | Drives playwright against the running app; checks every acceptance item from spec.md; dispatches fixer subagents on failures | spec.md + running app | verify.log.md |
| 5 | visual-qa-from-design | Screenshots each view, compares multimodally against the original design source, dispatches visual-fixer subagents for pixel-level gaps | spec.md design source + running app | visual-qa.md |
All files land under docs/design-to-code/<YYYY-MM-DD>-<topic>/ in your project:
docs/design-to-code/2025-06-01-cart-empty-state/
├── spec.md ← stage 1 (user-approved; never edited by assistant afterward)
├── plan.md ← stage 2
├── progress.md ← stage 3 (one entry per task)
├── verify.log.md ← stage 4 (one entry per acceptance item)
├── visual-qa.md ← stage 5 (one entry per view)
└── screenshots/ ← stage 5 screenshots
No skill writes into another skill's artifact.
npx claudepluginhub zhangtianwei1998/design-to-code --plugin design-to-codeTools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Browser automation and end-to-end testing MCP server by Microsoft. Enables Claude to interact with web pages, take screenshots, fill forms, click elements, and perform automated browser testing workflows.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses