@vincikit/mcp-server

@vincikit/mcp-server

AI stack configurator that helps you build the optimal technology stack through conversational interviews and intelligent recommendations.

What is VinciKit?

VinciKit is an MCP server that acts as your personal AI stack architect. Through a conversational interview, it understands your project's needs and recommends the best combination of AI models, tools, and services - complete with cost estimates and implementation guidance.

Features

• Conversational Interview - Guided questions to understand your project scope, expertise, budget, and timeline
• AI-Powered Recommendations - Intelligent stack suggestions using Claude or Gemini
• Tool Catalog - 30+ tools across 13 categories (finance, design, productivity, e-commerce, hardware, and more)
• Cost Estimation - Setup costs, monthly costs, and first-year projections
• Session Persistence - Save and resume sessions across devices
• ASCII Visualization - Text-based stack diagrams for any terminal

Installation

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "vincikit": {
      "command": "npx",
      "args": ["-y", "@vincikit/mcp-server"],
      "env": {
        "ANTHROPIC_API_KEY": "your-key-here"
      }
    }
  }
}

Global Install

npm install -g @vincikit/mcp-server
vincikit-mcp

From Source

git clone https://github.com/idoco2003/VinciKit.git
cd VinciKit/mcp-server
npm install
npm start

Available Tools

Interview Flow

Tool	Description
start_interview	Begin a new conversational stack configuration session
answer_question	Answer the current interview question
skip_to_recommendation	Jump directly to AI recommendations

Session Management

Tool	Description
get_session	Retrieve current session data and progress
update_session	Modify answers and regenerate recommendations
list_sessions	View all sessions for a user
link_account	Link session to email for persistence
create_setup_link	Generate a personalized setup URL

Catalog & Visualization

Tool	Description
list_tools	Browse the tool catalog by category
get_tool_details	Get detailed information about a specific tool
visualize_stack	Generate ASCII visualization of your stack
quick_recommend	Get fast recommendations without full interview

Resources

URI	Description
vincikit://catalog	Complete tool catalog (JSON)
vincikit://questions	Interview question definitions (JSON)

Tool Categories

• Finance - Alpha Vantage, Alpaca, Plaid, CoinGecko, Yahoo Finance
• Design/CAD - AutoCAD, Fusion 360, Figma, Canva
• Productivity - Notion, Google Sheets, Slack, GitHub
• E-commerce - Shopify, Square, Stripe
• Hardware - DigiKey, Mouser
• Maps/Location - Google Maps, Mapbox
• Data/Search - Brave Search, Semantic Scholar
• Food/Agriculture - USDA FoodData Central
• Transcription - AssemblyAI, Deepgram
• Edge AI - Jetson Orin Nano/NX, Raspberry Pi
• Vision - OpenCV, YOLO

Environment Variables

Variable	Required	Description
ANTHROPIC_API_KEY	Yes*	Claude API key for recommendations
GEMINI_API_KEY	Yes*	Gemini API key (alternative to Claude)
VINCIKIT_API_URL	No	Custom API endpoint
MCP_API_KEY	No	API key for remote access

*At least one AI provider key is required

Transport Modes

stdio (default)

For local Claude Desktop integration:

npm start

HTTP/SSE

For remote deployment (Cloud Run, etc.):

npm run start:http

Example Usage

User: I want to build a smart inventory system for my small retail business

AI: [calls start_interview]
    Let me help you find the perfect stack! First question:
    What's the scale of this project?
    - Hobby/Learning
    - Venture/Startup
    - Scale/Growth
    - Moonshot

User: Venture

AI: [calls answer_question with "venture"]
    Great! What's your background?
    ...

[After completing interview]

AI: [displays recommendation]
    Recommended Stack:
    - AI Engine: Claude 3.5 Sonnet ($20/month)
    - Tools: Shopify, Google Sheets, Notion
    - Estimated first year cost: $1,200

Links

• Website: https://vincikit.com
• Repository: https://github.com/idoco2003/VinciKit

License

MIT