sb-figma
starsforks
How to install sb-figma?
strongeron/storybook-workbench/sb-figma
$npx -y skills add strongeron/storybook-workbench --skill sb-figmaThe Figma↔Storybook bridge, both directions, via the native Figma MCP. design→code: map foundation tokens (color/spacing/type) from Figma variables with design↔code parity + drift, and deliver approved Figma components (extract → build → embed), authoring stories via sb-stories'
SKILL.md
| 1 | # sb-figma — Figma → production Storybook delivery |
| 2 | |
| 3 | `sb-figma` owns the **DELIVER** stage of the Figma→Storybook lifecycle: an **approved** design becomes |
| 4 | production Storybook. It is Figma-aware (uses the Figma MCP), but it is **not** the exploration skill. |
| 5 | |
| 6 | > **The lifecycle line (load this first).** *Undecided / trying options* → **`sb-explore`** (Lab sandbox, |
| 7 | > iterate against a Figma node) → **`sb-ship`** (graduate the Lab experiment). *Already approved in Figma* → |
| 8 | > **`sb-figma`** (deliver direct to prod). Both touch Figma MCP; the difference is **exploration vs delivery**. |
| 9 | > If the user is still deciding, hand off to `sb-explore`. See `references/figma-token-sync.md`. |
| 10 | |
| 11 | ## Job 0 — capture the MCP output (always, before anything else) |
| 12 | |
| 13 | > **Force the NATIVE structured tools — never work from a screenshot.** A screenshot is pixels; it cannot |
| 14 | > give you variables, styles, or component props. The full picture comes ONLY from the native MCP tools, and |
| 15 | > you must pull all three categories before building or connecting: |
| 16 | > - **Variables** → `get_variable_defs` (the token values: color/spacing/type/effect, resolved). |
| 17 | > - **Components + styles** → `get_design_context` (the reference code, applied styles, props/variants). |
| 18 | > - **Structure** → `get_metadata` (the node tree; on truncation, drill to child node-ids — never give up at |
| 19 | > the parent). |
| 20 | > `get_screenshot` is **visual reference only** — for an eyeball diff after you've built from the structured |
| 21 | > data. NEVER read tokens, props, or layout off a screenshot. If a tool returns "nothing selected" on a page |
| 22 | > id, drill to a concrete component node (a page is not a layer). |
| 23 | |
| 24 | Scripts can't call the Figma MCP, and an MCP result lives only in the agent's context — ephemeral, gone |
| 25 | on the next session, absent headless. So **every Figma MCP call you make, persist it** through the universal |
| 26 | store before using it. This is what makes the pipeline reproducible and iterable. |
| 27 | |
| 28 | ```bash |
| 29 | # any JSON-returning tool — pipe its output straight in: |
| 30 | <get_variable_defs output> | node scripts/capture-figma.mjs --tool get_variable_defs --file <FILE> --node <NODE> --from-mcp - |
| 31 | <get_design_context output> | node scripts/capture-figma.mjs --tool get_design_context --file <FILE> --node <NODE> --from-mcp - |
| 32 | <get_metadata output> | node scripts/capture-figma.mjs --tool get_metadata --file <FILE> --node <NODE> --from-mcp - |
| 33 | <get_code_connect_map out> | node scripts/capture-figma.mjs --tool get_code_connect_map --file <FILE> --node <NODE> --from-mcp - |
| 34 | # get_screenshot returns an image — save it, then register the file: |
| 35 | node scripts/capture-figma.mjs --tool get_screenshot --file <FILE> --node <NODE> --image /tmp/frame.png |
| 36 | # see the whole inventory (degrade / iterate): |
| 37 | node scripts/capture-figma.mjs --list |
| 38 | ``` |
| 39 | |
| 40 | Store layout: `.storybook/figma/manifest.json` + `.storybook/figma/<tool>/<node>.json` (images keep their |
| 41 | ext). Re-capturing a (tool,node) overwrites — diff against git to see what moved in Figma. **Downstream steps |
| 42 | read the store, never re-call MCP.** |
| 43 | |
| 44 | ## MCP realities (field-verified against the Detections file, 2026-06-22) |
| 45 | |
| 46 | What the live Figma MCP actually returns — the scripts already handle these; know them so you don't fight the output: |
| 47 | - **`get_variable_defs` is a FLAT `{ "name": "value" }` map** (not nested DTCG, not an array). Colors come **already |
| 48 | resolved to hex** (`"semantic/background":"#fbfcfc"`), incl. 8-digit alpha (`"#e4e5e580"`). Numbers are bare |
| 49 | strings (`"spacing-2":"8"`, `"wght/semibold":"650"`). **Ty |
People who liked this also viewed
6 picksType
- web-design-guidelinesReview UI code for Web Interface Guidelines compliance. Use when asked to "review my UI", "check accessibility", "audit design", "review UX", or "check my site against best practices".SkillsJul 2026
- lark-slides飞书幻灯片：创建和编辑幻灯片。创建演示文稿、读取幻灯片内容、管理幻灯片页面（创建、删除、读取、局部替换）。当用户需要创建或编辑幻灯片、读取或修改单个页面时使用。当用户给出 doubao.com 的 /slides/ URL/token 时，也应直接使用本 skill，不要因为域名不是飞书而回退到 WebFetch；路由依据是 URL 路径模式和 token，而不是域名。不负责：云文档内容编辑（走 SkillsJul 2026
- design-taste-frontendTaste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop SkillsJul 2026
- ui-ux-pro-maxUI/UX design intelligence for web and mobile. Searchable local database with 84 styles, 192 color palettes, 74 font pairings, 192 product types, 98 UX guidelines, 104 icon entries, 16 GSAP motion presSkillsJul 2026
- caveman-helpQuick-reference card for all caveman modes, skills, and commands. One-shot display, not a persistent mode. Trigger: /caveman-help, "caveman help", "what caveman commands", "how do I use caveman".SkillsJul 2026
- codebase-designShared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-nSkillsJul 2026
