neuria-agent
NEURIA - AI-powered Awwwards-level website generator with WebGL/Three.js
Documentation
NEURIA — Emotion-Driven Design Intelligence for AI Coding Tools
NEURIA is an MCP (Model Context Protocol) server that gives AI assistants like Claude, Cursor, and Windsurf a *sense of design taste*. Instead of relying on generic Tailwind defaults or hardcoded color palettes, AI agents query NEURIA for emotion-aligned design tokens — colors, typography, spacing, animations — all backed by real-world data from 300+ award-winning websites.
> "Design is not just what it looks like. Design is how it feels." — Steve Jobs
NEURIA quantifies that *feeling* into actionable design tokens.
What is NEURIA?
NEURIA maps 6 core emotions to complete design systems:
| Emotion | Vibe | Example Brands |
|---|---|---|
| Trust | Stable, reliable, professional | Linear, Stripe, Notion |
| Excitement | Bold, energetic, dynamic | Vercel, Arc, Framer |
| Calm | Serene, minimal, breathable | Headspace, Muji, Apple |
| Luxury | Premium, refined, exclusive | Porsche, Rolex, Aesop |
| Innovation | Futuristic, cutting-edge, smart | OpenAI, Tesla, Figma |
| Energy | Vibrant, playful, high-tempo | Spotify, Discord, Duolingo |
Each emotion includes color palettes, typography settings, spacing scales, border radii, shadow styles, and animation parameters — all derived from statistical analysis of real CSS extracted from 300+ top websites.
Quick Start
1. Add to your MCP config
Add to your `.mcp.json` (Claude Code) or IDE MCP settings:
{
"mcpServers": {
"neuria": {
"type": "sse",
"url": "https://api.neuria.tech/mcp"
}
}
}That's it. No API key needed for core tools.
2. Local Development (Optional)
git clone https://github.com/kosukeuemura0220-del/neuria-agent.git
cd neuria-agent
pip install -r requirements.txt
cp .env.example .env
# Fill in API keys for advanced tools (see .env.example)
python3 mcp-servers/neuria/server.py10 MCP Tools
Core Tools (No API Key Required)
| # | Tool | Description | Input |
|---|---|---|---|
| 1 | `get_emotion_tokens` | Get CSS custom properties for an emotion | `emotion`, `category` |
| 2 | `search_similar_designs` | Find similar designs via pgvector cosine similarity | `query`, `mood`, `component_type` |
| 3 | `get_color_palette` | Get a 5-color palette + Color API scheme | `emotion`, `industry` |
| 4 | `get_typography` | Get font settings for emotion × use case | `emotion`, `use_case` |
| 5 | `score_html` | Score HTML against NEURIA design standards | `html_content`, `target_emotion`, `url` |
| 8 | `get_animation_tokens` | Get GSAP / CSS / Spring animation parameters | `emotion`, `context` |
Advanced Tools (API Keys Required)
| # | Tool | Description | Required Keys |
|---|---|---|---|
| 6 | `generate_hero_image` | AI-generated emotion-matched hero image | 3D AI Studio |
| 7 | `analyze_reference_url` | Extract design tokens from any URL (W3C DTCG format) | `FIRECRAWL_API_KEY` |
| 9 | `analyze_visual_design` | Dual-analysis: screenshot colors + CSS analysis | `SCREENSHOTONE_API_KEY`, `IMAGGA_API_KEY` |
| 10 | `analyze_brand_voice` | Text emotion analysis via Hume AI | `HUME_API_KEY` |
Example Usage
1. Generate a Landing Page with Emotion Tokens
User: "Analyze linear.app and create a SaaS landing page with the same design feel"
→ NEURIA calls: analyze_reference_url("https://linear.app")
Returns: dominant_emotion="innovation", W3C DTCG design tokens, emotion_scores
→ NEURIA calls: get_emotion_tokens("innovation")
Returns: CSS variables (--neuria-bg, --neuria-accent, --neuria-font-*, etc.)
→ NEURIA calls: get_animation_tokens("innovation", "page_load")
Returns: GSAP ease, CSS cubic-bezier, Spring physics params
→ Claude generates HTML using NEURIA tokens (no hardcoded colors)
→ NEURIA calls: score_html(generated_html, "innovation")
Returns: composite_score=87/100 ✅2. Quick Color Palette for Fintech
User: "I need a trustworthy color palette for a fintech app"
→ get_color_palette("trust", industry="fintech")
Returns:
{
"background": "#0B1120",
"surface": "#131D36",
"accent": "#2E7DFF",
"text_primary": "#E8ECF4",
"text_secondary": "#7A8BAB",
"color_scheme": [...5 harmonious colors from The Color API]
}3. Brand Voice Analysis
User: "What emotion does Apple's copy convey?"
→ analyze_brand_voice(url="https://apple.com")
Returns:
{
"dominant_emotion": "innovation",
"neuria_emotions": {
"innovation": 0.31,
"luxury": 0.24,
"calm": 0.19,
...
}
}Architecture
┌──────────────────────────────────────────────┐
│ AI Coding Tool │
│ (Claude Code / Cursor / Windsurf) │
└──────────────┬───────────────────────────────┘
│ MCP Protocol (stdio / SSE)
▼
┌──────────────────────────────────────────────┐
│ NEURIA MCP Server │
│ │
│ ┌─────────┐ ┌──────────┐ ┌──────────────┐ │
│ │ Emotion │ │ Supabase │ │ External │ │
│ │ Token DB │ │ pgvector │ │ APIs │ │
│ │ (6×CSS) │ │ (300+ │ │ │ │
│ │ │ │ sites) │ │ • Firecrawl │ │
│ └─────────┘ └──────────┘ │ • Color API │ │
│ │ • ScreenShot │ │
│ ┌─────────────────────┐ │ • Imagga │ │
│ │ Animation Tokens │ │ • Hume AI │ │
│ │ (6 emotions × │ │ • WAVE │ │
│ │ 4 contexts) │ │ • 3D AI │ │
│ └─────────────────────┘ └──────────────┘ │
└──────────────────────────────────────────────┘Scoring System
`score_html` evaluates generated HTML across 4 dimensions:
| Dimension | Weight | What it Measures |
|---|---|---|
| Aesthetic | 35% | Visual harmony, whitespace, color usage |
| Emotion | 35% | Alignment with target emotion tokens |
| Technical | 15% | Clean markup, performance, responsive |
| Accessibility | 15% | WCAG 2.1 AA (lang, alt, contrast, headings) |
Target: 85+ / 100 for production-quality output.
Environment Variables
Create a `.env` file at the project root (see `.env.example`):
# Required for core functionality
SUPABASE_URL=your_supabase_url
SUPABASE_KEY=your_supabase_anon_key
NEURIA_API_BASE=your_api_base_url
# Required for analyze_reference_url
FIRECRAWL_API_KEY=your_firecrawl_key
# Required for generate_hero_image
THREE_D_AI_STUDIO_KEY=your_3d_ai_key
# Required for analyze_visual_design
SCREENSHOTONE_API_KEY=your_key
IMAGGA_API_KEY=your_key
IMAGGA_API_SECRET=your_secret
# Required for analyze_brand_voice
HUME_API_KEY=your_hume_key
# Optional — WAVE accessibility API
WAVE_API_KEY=your_wave_keyPricing
> Coming soon
| Plan | Price | Requests |
|---|---|---|
| Free | $0/mo | 5 req/day |
| Indie | $19/mo | 500 req/mo |
| Pro | $49/mo | 2,000 req/mo |
| Team | $149/mo | 10,000 req/mo |
| Enterprise | Custom | Unlimited |
Links
- Website: https://neuria.tech
- MCP Endpoint: `https://api.neuria.tech/mcp`
- API Docs: https://api.neuria.tech/docs
- GitHub: https://github.com/kosukeuemura0220-del/neuria-agent
License
MIT License — see LICENSE for details.
NEURIA — Because AI-generated UI shouldn't look like AI-generated UI.
Frequently asked questions
What is neuria-agent?
neuria-agent is NEURIA - AI-powered Awwwards-level website generator with WebGL/Three.js
How do I install neuria-agent?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is neuria-agent open source?
Yes — it is hosted on GitHub at https://github.com/kosukeuemura0220-del/neuria-agent.
Related MCP tools
Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory across sessions with a self-hosted knowledge graph engine.
Automate browser based workflows with AI
Hindsight: Agent Memory That Learns
A privacy-first app that strips AI watermarks from content you own.
Agent framework and applications built upon Qwen>=3.0, featuring Function Calling, MCP, Code Interpreter, RAG, Chrome extension, etc.
The power of Claude Code / GeminiCLI / CodexCLI + [Gemini / OpenAI / OpenRouter / Azure / Grok / Ollama / Custom Model / All Of The Above] working as one.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP