The article emphasizes the importance of providing clear context and structured data when using AI to generate production-ready code from design systems, particularly within Figma. A recent Figma report highlights that while 68% of developers utilize AI for coding, only 32% trust the outputs. The primary issue is the AI's ability to understand context rather than its coding capability. To bridge this gap, designers need to supply clean data and detailed prompts. The Model Context Protocol (MCP) facilitates this by acting as an intermediary that enriches Figma files with meaningful information for AI. Key strategies include using semantic tokens instead of basic color names, which helps AI understand the purpose behind design choices. Features like Figma's new slots allow for flexible content areas, enhancing how designs translate to code. Annotation tools in Figma can further clarify details like interaction requirements, ensuring that the AI has the necessary context. The article also suggests defining components as structured data, such as JSON, rather than mere visual representations in Figma. This approach aligns with the needs of AI, allowing for more accurate interpretations of designs. To guide AI behavior effectively, the article recommends creating a "cheat sheet" with custom rules. This involves establishing foundational rules, design system guidelines, and Figma MCP workflow specifics. By organizing these rules in a dedicated folder within the project, developers can reference them consistently, enhancing the AI's understanding and output quality. Overall, this structured approach not only improves communication between design and code but also boosts efficiency in the development process.