Reftrix

About

Reftrix is a sophisticated platform designed to centralize and analyze web design patterns, encompassing layout structures, motion effects, and overall quality. It leverages advanced vector search (pgvector) and RAG (Retrieval Augmented Generation) techniques to create a searchable knowledge base, allowing users to discover intricate design elements efficiently. Integrated with Claude via 19 specialized MCP tools, Reftrix empowers designers and developers with AI-driven insights for detailed page analysis, semantic pattern discovery, and multi-dimensional quality assessments, streamlining the evaluation and inspiration process for web projects.

Key Features

• 0 GitHub stars
• Detect CSS/JS animations, scroll-triggered effects, and motion patterns.
• Perform 768-dimensional vector search with HNSW indexing for design pattern discovery.
• Extract and analyze page layout structures, section patterns, and design components.
• Utilize Reciprocal Rank Fusion for hybrid search combining vector (60%) and full-text (40%) capabilities.
• Evaluate design quality across multiple dimensions (originality, craftsmanship, contextuality).

Use Cases

• Discovering and analyzing web design patterns for inspiration and best practices.
• Evaluating the quality and originality of web designs across various metrics.
• Conducting comprehensive analysis of single URLs for layout, motion, quality, and narrative.