Soma Nexus MCP
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Soma Nexus MCP fetch the file tree from my Figma project"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Soma Nexus MCP
MCP server em TypeScript para VS Code, focado em leitura de dados do Figma e contexto para design-to-code.
O que faz
Busca árvore de arquivo do Figma
Busca nós específicos por ID
Resolve URL completa do Figma (
fileKey+node-id)Busca nó direto a partir de URL
Retorna saída normalizada para design-to-code
Busca imagens exportadas
Busca components, styles e variables
Valida contrato federado de mapeamento Figma -> Design System
Related MCP server: Figma MCP
Arquitetura (simples e prática)
Aplicamos SOLID + Clean Architecture sem excesso de abstração:
src/index.ts: composition rootsrc/presentation/register-tools.ts: camada de entrada (tools MCP)src/infrastructure/figma-api-client.ts: integração com API Figmasrc/domain/figma-url.ts: regras de URL e IDssrc/application/node-normalizer.ts: normalização para design-to-codesrc/core/errors.tsesrc/core/responses.ts: erros e contrato de resposta
Regras usadas:
Responsabilidade única por módulo
Dependências apontando para dentro (presentation -> application/domain/infrastructure)
Infra isolada (troca de API client sem mexer em tools)
Erro e resposta padronizados para todas as tools
Requisitos
Node.js 18+
FIGMA_TOKENdefinido no ambiente do cliente MCPOpcional:
FIGMA_NODES_CACHE_TTL_MS(default:90000)
Desenvolvimento local
npm install
npm run build
npm run dev
npm testTools disponíveis
get_figma_file: payload bruto do arquivo inteiro (debug pesado)parse_figma_url: extraifileKeyenodeIdde URLget_figma_nodes: busca nós por IDsget_figma_node_from_url: busca nó direto com URL completaget_figma_node_normalized: metadata + estilos/tokens/assets normalizadosget_figma_images: URLs de renderização de nósget_figma_components: componentes publicadosget_figma_styles: estilos publicadosget_figma_variables: variáveis locaisvalidate_figma_to_code_index: valida índice global e docs por componentepreview_figma_component_mapping: mostra melhor match de componente DS para um node Figmagenerate_flutter_component_from_figma: gera snippet Flutter mínimo via mapping federadogenerate_flutter_page_from_figma: gera estrutura Flutter de página inteira com DS quando houver match
Economia de tokens (recomendado)
Use estes parâmetros em tools de leitura/mapping/geração:
responseMode:compact|standarddebugTelemetry:true|false
Sugestão para uso normal:
responseMode: "compact"debugTelemetry: false
Quando quiser medir custo real de payload:
debugTelemetry: trueretornameta.payloadSizeBytesemeta.responseSizeBytes.
Observações:
Tools com
depthusam default conservador1quando não informado.Preview de matching limita candidatos por padrão para reduzir resposta (
compact: 3,standard: 10).
Modelo federado Figma -> Code
Cada componente do Design System deve manter seu próprio arquivo:
{component}/docs/figma-to-code.md
E todos os componentes devem ser registrados em:
docs/figma-to-code-index.yaml
Documentação de suporte:
docs/specs/figma-to-code-component-doc-spec.mddocs/figma-to-code-authoring-guide.mddocs/templates/component-figma-to-code.template.md
Configuração no VS Code
No workspace, use .vscode/mcp.json:
{
"inputs": [
{
"id": "figma-token",
"type": "promptString",
"description": "FIGMA_TOKEN for Soma Nexus MCP",
"password": true
}
],
"servers": {
"soma-nexus-figma-mcp": {
"type": "stdio",
"command": "npm",
"args": ["run", "dev"],
"env": {
"FIGMA_TOKEN": "${input:figma-token}"
}
}
}
}Na primeira vez, VS Code pergunta token e guarda valor localmente.
Fluxo recomendado (design-to-code)
validate_figma_to_code_indexparse_figma_urlpreview_figma_component_mappinggenerate_flutter_component_from_figma(quando URL aponta para componente)generate_flutter_page_from_figma(quando URL aponta para frame/página)get_figma_node_normalized(debug/inspeção)get_figma_images(se precisar assets)
Esse fluxo evita payload gigante e melhora estabilidade da geração de código.
Contrato de resposta
Sucesso:
{
"ok": true,
"data": {}
}Erro:
{
"ok": false,
"error": {
"code": "FIGMA_UNAUTHORIZED",
"message": "Figma token unauthorized.",
"details": {}
}
}Available Tools
13 toolsgenerate_flutter_component_from_figmaBRead-only
Generate a minimal Flutter widget snippet from a Figma node using federated docs mapping.
Inputs: figmaUrl (required), depth, version, responseMode, +1 more.
generate_flutter_page_from_figmaBRead-only
Generate Flutter page scaffold from a Figma page/frame URL, using DS components when matched.
Inputs: figmaUrl (required), depth, version, maxNodes, +2 more.
get_figma_componentsARead-only
Fetch published components found in a Figma file.
Inputs: fileKey (required), responseMode, debugTelemetry.
get_figma_fileBRead-only
Fetch a Figma file. Prefer compact mode to avoid large payloads.
Inputs: fileKey (required), version, responseMode, debugTelemetry.
get_figma_imagesBRead-only
Render Figma nodes to image URLs for assets and previews.
Inputs: fileKey (required), nodeIds (required), scale, format, +3 more.
get_figma_node_from_urlARead-only
Fetch a specific Figma node directly from a full Figma URL.
Inputs: figmaUrl (required), depth, version, geometry, +2 more.
get_figma_node_normalizedARead-only
Fetch a Figma node from URL and return normalized metadata, tokens, styles, and asset refs.
Inputs: figmaUrl (required), depth, version, responseMode, +2 more.
get_figma_nodesBRead-only
Fetch specific Figma nodes, their subtrees, and optional geometry.
Inputs: fileKey (required), nodeIds (required), depth, version, +3 more.
get_figma_stylesCRead-only
Fetch published styles found in a Figma file.
Inputs: fileKey (required), responseMode, debugTelemetry.
get_figma_variablesBRead-only
Fetch local variables and collections from a Figma file.
Inputs: fileKey (required), responseMode, debugTelemetry.
parse_figma_urlARead-only
Parse a Figma URL and extract fileKey and nodeId.
Inputs: figmaUrl (required).
preview_figma_component_mappingBRead-only
Preview how a Figma node matches a Design System component via federated docs.
Inputs: figmaUrl (required), depth, version, responseMode, +2 more.
validate_figma_to_code_indexARead-only
Validate the global figma-to-code index and all per-component docs before generation.
Inputs: responseMode, debugTelemetry.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 13 tools
Multiple tools have overlapping boundaries, especially the node-fetching trio: get_figma_nodes, get_figma_node_from_url, and get_figma_node_normalized all retrieve node data with subtle differences. get_figma_node_from_url and get_figma_node_normalized are particularly confusing since both take a URL and return node information, with the latter adding normalization. This overlap makes it difficult for an agent to reliably select the right tool.
Most tools follow a verb_noun pattern (get_figma_file, generate_flutter_component_from_figma), but there are inconsistencies in granularity and style. For example, get_figma_nodes (plural) coexists with get_figma_node_from_url and get_figma_node_normalized (singular), and the latter two are not parallel to each other. The mix of 'from_url' and 'normalized' suffixes adds unpredictability, so while the convention is readable, it is not consistently applied.
With 13 tools, the server is well-scoped for a Figma-to-Flutter generation workflow. Each tool addresses a distinct step in the pipeline (fetching, parsing, mapping, validating, generating) without unnecessary bloat. The count fits comfortably within the ideal 3-15 range and each tool appears to serve a concrete purpose.
The tool set covers the core lifecycle of Figma-to-Flutter generation: fetching files/nodes/styles/variables, parsing URLs, validating index, previewing component mappings, and generating both component and page code. Minor gaps exist, such as no explicit tool to manage or update the component index or design system docs, but agents can work around these with the existing validation and mapping tools.
Related MCP Connectors
The Figma MCP server brings Figma design context directly into your AI workflow.
MCP server for interacting with the Supabase platform
MCP server for Statsig API - interact with Statsig's feature flags, experiments, and analytics
MCP server for Financial Modeling Prep API with 250+ financial data tools
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceRead-only Figma MCP server that enables design-to-code workflows by talking to the Figma REST API with a personal access token, for use with Claude Code and GitHub Copilot.2,146 npmMIT
- FlicenseNot gradedqualityDmaintenanceMCP server for the Figma API. Lets AI agents fetch designs, nodes, and rendered images from Figma.2,146 npm-
- FlicenseNot gradedqualityDmaintenanceRead-only MCP server that wraps the Figma REST API, providing tools like get file metadata, list frames, and export node image URLs for AI SDLC agents.-
- AlicenseBqualityCmaintenanceLocal MCP server connecting AI clients to the Figma desktop app for inspecting and editing Figma documents via the Plugin API.1616 npm1MIT