zimai233/dsh-figma-to-lottie

为 DeepSeek Harness 打造的 Figma/SVG → Lottie 动画编译器。

将 SVG 路径与关键帧数据编译成自包含的 .lottie.json 文件——直接在 agent 对话中完成，无需任何设计工具。

为什么需要它

DSH 插件生态没有设计工具桥接层（虽然有 Figma MCP，但没有插件能把设计数据编译成动画资产）。本插件填补了这个空白：它把 SVG→Lottie 的编译管道——路径解析、贝塞尔进出切线转换、关键帧缓动——提炼成两个可被 agent 调用的工具。

功能特性

• lottie_compile_shape — 将 SVG 路径 d 字符串解析为 Lottie 形状值 { i, o, v, c }，带正确的贝塞尔切线。编译完整动画前，可先检查任意 SVG 路径的转换结果。
• lottie_compile — 从紧凑的图层规格组装完整的 Lottie 5.7.4 JSON 文档：矩形（支持渐变填充）、路径、图片（base64 内嵌）、文字，每层均支持透明度 / 上滑关键帧动画。
• 零构建步骤——纯 ESM，发布即运行时代码。无需 prepare 脚本，无需构建授权。

安装

# 从 npm 安装
dsh plugin --profile myprofile add dsh-figma-to-lottie

# 从 GitHub 安装（建议锁定 commit，供应链更安全）
dsh plugin --profile myprofile add github:zimai233/dsh-figma-to-lottie#<sha>

使用

用自然语言告诉 agent：

"编译一个 3 秒的加载动画：345×414 画布，浅灰背景，10 根蓝色柱子在 x=215.5 处以 0.48/0.24/0.4/1.4 的缓动上滑，每根错开 100ms。写入 ./loading.lottie.json"

agent 会调用 lottie_compile，传入类似这样的规格：

{
  "frame": { "w": 345, "h": 414, "bg": "#f3f4f6", "name": "loading" },
  "fps": 60,
  "durationMs": 3000,
  "layers": [
    { "type": "rect", "id": "bar-1", "x": 215.5, "y": 112, "w": 57, "h": 13, "r": 6.5,
      "gradient": ["#61adf8", "#8fd1fa"], "slideUp": 15, "startMs": 0 },
    { "type": "rect", "id": "bar-2", "x": 215.5, "y": 137, "w": 57, "h": 13, "r": 6.5,
      "gradient": ["#61adf8", "#8fd1fa"], "slideUp": 15, "startMs": 100 }
  ]
}

产出的 Lottie JSON 可直接用于 iOS / Android / Web（LottieWeb、lottie-ios、lottie-android）。

图层规格参考

字段	类型	说明
frame.w / frame.h	number	画布尺寸（像素，必填）
frame.bg	hex 字符串	可选背景色
fps	number	帧率，默认 60
durationMs	number	动画总时长（必填）
bezier	{p1x,p1y,p2x,p2y}	可选缓动曲线，默认为上滑缓动

图层类型

类型	字段
rect	x、y、w、h、r?、fill?、gradient?: [c1,c2]、slideUp?、opacity?、startMs?、endMs?
path	d（SVG 路径）、x?、y?、fill?、opacity?
image	path（绝对文件路径，PNG 以 base64 内嵌）、x、y、w、h
text	text、x、y、size、color?、font?

开发

npm install
npm test          # 对 buildLottie + svgPathToLottie 运行纯核心测试
npm run pack      # 打包预检（相当于 pnpm pack）

许可证

MIT

---

dsh-figma-to-lottie

Figma/SVG → Lottie animation compiler for DeepSeek Harness.

Turn SVG paths and keyframe data into self-contained .lottie.json files — directly from your agent conversation, no design tool required.

Why

The DSH plugin ecosystem has no design-tool bridge (the Figma MCP exists, but no plugin compiles design data into animation assets). This plugin fills that gap: it extracts the SVG→Lottie compilation pipeline — path parsing, bezier in/out tangent conversion, keyframe easing — into two agent-callable tools.

Features

• lottie_compile_shape — parse an SVG path d string into a Lottie shape value { i, o, v, c } with correct bezier tangents. Inspect how any SVG path translates before compiling a full animation.
• lottie_compile — assemble a complete Lottie 5.7.4 JSON document from a compact layer spec: rects (with gradient fills), paths, images (base64-embedded), and text, each with per-layer opacity/slide-up keyframe animation.
• Zero build step — pure ESM, the published package is the runtime code. No prepare script, no build permission needed.

Install

# from npm
dsh plugin --profile myprofile add dsh-figma-to-lottie

# from GitHub (lock the commit for supply-chain hygiene)
dsh plugin --profile myprofile add github:zimai233/dsh-figma-to-lottie#<sha>

Usage

Ask the agent in natural language:

"Compile a 3-second loading animation: 345×414 frame, light gray background, 10 blue bars at x=215.5 that slide up with a 0.48/0.24/0.4/1.4 ease, staggered 100ms apart. Write it to ./loading.lottie.json"

The agent calls lottie_compile with a spec like:

{
  "frame": { "w": 345, "h": 414, "bg": "#f3f4f6", "name": "loading" },
  "fps": 60,
  "durationMs": 3000,
  "layers": [
    { "type": "rect", "id": "bar-1", "x": 215.5, "y": 112, "w": 57, "h": 13, "r": 6.5,
      "gradient": ["#61adf8", "#8fd1fa"], "slideUp": 15, "startMs": 0 },
    { "type": "rect", "id": "bar-2", "x": 215.5, "y": 137, "w": 57, "h": 13, "r": 6.5,
      "gradient": ["#61adf8", "#8fd1fa"], "slideUp": 15, "startMs": 100 }
  ]
}

The result is a ready-to-play Lottie JSON for iOS / Android / Web (LottieWeb, lottie-ios, lottie-android).

Layer Spec Reference

Field	Type	Description
frame.w / frame.h	number	Canvas size in px (required)
frame.bg	hex string	Optional background fill
fps	number	Frames per second, default 60
durationMs	number	Total animation length (required)
bezier	{p1x,p1y,p2x,p2y}	Optional easing curve, defaults to slide-up ease

Layer types

Type	Fields
rect	x, y, w, h, r?, fill?, gradient?: [c1,c2], slideUp?, opacity?, startMs?, endMs?
path	d (SVG path), x?, y?, fill?, opacity?
image	path (absolute file path, PNG base64-embedded), x, y, w, h
text	text, x, y, size, color?, font?

Development

npm install
npm test          # runs the pure-core tests against buildLottie + svgPathToLottie
npm run pack      # dry-run publish (pnpm pack equivalent)

License

MIT