NEURIA Design Intelligence MCP

What it is

An MCP server you run locally that maps 6 emotions — trust, excitement, calm, luxury, innovation and energy — onto complete design systems, derived from CSS collected across 300+ award-winning sites. An assistant queries it for tokens instead of inventing hex codes.

What you get

• Tools that need no key, built on data bundled in the server: get_emotion_tokens for CSS custom properties, get_typography, get_animation_tokens for GSAP, CSS and spring parameters, and get_color_palette, a five-colour palette extended with a matching scheme from The Color API
• score_html grades generated markup against a target emotion across aesthetic, emotion, technical and accessibility dimensions and returns a composite score; search_similar_designs finds comparable designs in a design database
• Tools that need third-party keys: analyze_reference_url extracts tokens from any URL in W3C DTCG format, analyze_visual_design combines screenshot colour analysis with CSS, analyze_brand_voice reads emotion out of text, and generate_hero_image produces a matched image
• 10 tools in total

Requirements

Python 3.10 or later. Clone the repository, pip install -r requirements.txt, then run python3 mcp-servers/neuria/server.py, which serves SSE at http://127.0.0.1:8000/sse; the hosted endpoint in the README no longer resolves. The four token tools need no key; get_color_palette also reaches The Color API over the internet. search_similar_designs needs a Supabase database you build and fill yourself, reached through SUPABASE_URL and SUPABASE_SERVICE_KEY (the server reads that name, not the SUPABASE_KEY in .env.example); the repository's migrations do not create the search function it calls, and without a database it only ranks the other emotions by similarity. score_html needs the scoring API from pipeline/api_server.py running at NEURIA_API_BASE, with ANTHROPIC_API_KEY set; both servers default to port 8000, so move one. The four advanced tools each require your own third-party credentials: FIRECRAWL_API_KEY, SCREENSHOTONE_API_KEY with IMAGGA_API_KEY and IMAGGA_API_SECRET, HUME_API_KEY, and THREEDAI_API_KEY for 3D AI Studio. MIT licensed.

Setup effort

Build from source — git clone https://github.com/kosukeuemura0220-del/neuria-agent.git

Tools you get

10 tools named in the project's documentation.

Requires

Python 3.10 or later with pip

Optional

• ANTHROPIC_API_KEY in your environment
• SUPABASE_URL in your environment
• SUPABASE_SERVICE_KEY in your environment
• FIRECRAWL_API_KEY in your environment
• SCREENSHOTONE_API_KEY in your environment
• IMAGGA_API_KEY in your environment
• IMAGGA_API_SECRET in your environment
• HUME_API_KEY in your environment
• THREEDAI_API_KEY in your environment

Where it runs

All 3 documented clients run the server on your machine.