This paper presents Slides Agent, an intelligent agent that leverages large language models (LLMs) to automate presentation creation and analyze visual design quality. The system comprises two main components: an analysis module that uses multimodal LLM capabilities for visual recognition and classification of design errors across seven categories (alignment, layout, contrast, color, formatting, size, typography), and a generation module that applies Chain of Thought approach for planning presentation structure followed by structured output of slides. A key feature of the system is its ability to automatically generate executable Python code for fixing detected design issues through the Google Slides API. Preliminary evaluation demonstrates quality generation of presentations that match source material. The proposed approach demonstrates the potential of using LLMs for creative tasks that require both semantic understanding of content and adherence to visual design principles.