
Inspo launches as an open-source Model Context Protocol server that gives AI coding agents access to real-world design systems and production website references to fix generic UI code generation.
Created by developer Hassan El Mghari (@Nutlope), Inspo is an open-source Model Context Protocol (MCP) server designed to overcome the "taste problem" in AI coding agents such as Claude Code, Cursor, Windsurf, and Zed. Instead of defaulting to generic dark-mode palettes and repetitive layouts, agents can query Inspo's curated archive of over 2,300 pages across 832+ production websites captured on desktop and mobile. The server extracts real design systems—including type ramps, color palettes, CSS variables, and spacing scales organized into structured DESIGN.md files—along with 68 canonical UI components featuring copy-pasteable JSX. Available via a single npx command and supporting both stdio and HTTP transports, Inspo enables AI agents to inspect proven production architectures and implement polished, context-appropriate user interfaces.
While LLM coding agents have mastered frontend syntax and state management, their default design output consistently falls into generic "AI slop" without taste; giving agents native MCP tools to study real production design systems is the most effective approach yet to bridging that aesthetic gap.
- –Solves the aesthetic cold-start: Rather than generating stereotypical purple gradients and generic card grids, agents can reference actual DOM autopsies and extracted CSS variables to mirror high-craft frontend standards.
- –Seamless agent integration: Providing 15 MCP tools alongside unified macros like `recommend(brief)` allows agents to autonomously inspect typography, layouts, and responsive structures directly inside their standard terminal or IDE workflow.
- –High-leverage static catalogue: Shipping the dataset as a static seed eliminates runtime database dependencies while providing instant access to hundreds of real-world sites and reusable JSX patterns.
- –Curation longevity challenge: Because web aesthetic trends and frontend libraries shift rapidly, static reference repositories will require active ongoing maintenance and automated capture pipelines to avoid serving dated UI patterns.
DISCOVERED
1h ago
2026-09-18
PUBLISHED
1h ago
2026-09-18
RELEVANCE
AUTHOR
AI LABS