design-to-code

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). O fluxo exige passar o token da API Figma diretamente em um comando (node scripts/... fetch-figma "url" "figd_..."), o que incentiva inserir/mostrar o PAT verbatim em comandos/saídas geradas pelo agente, criando risco de exfiltração de segredos.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text can enter the LLM context via fetch-figma → executeFigmaAndImagesActions(...) which downloads and processes a user-supplied Figma document (public/third-party content) into process/processed.json, then structure-prompt/props-prompt/code-prompt read that processed data and embed it into the prompts sent to the agent; the workflow explicitly uses the fetched Figma text (component names/props) as prompt input.