MCP-1st-Birthday/sipify-mcp

🧑‍💻 Sipify: AI Creative Ads Agent

One URL. Infinite Ads. Sipify is an AI Ad Creative Agent that understands your brand deeply — it analyzes your website, learns your audience segments, creates targeted Ad Personas, and finally generates high-performance Ad creatives personalized for each audience group. From strategy to final visuals, it designs ads that work for your brand.

🎥 Preview

<a href="https://youtu.be/9ICdPjvXJbU"> <img src="https://img.youtube.com/vi/9ICdPjvXJbU/maxresdefault.jpg" width="100%" alt="Sipify-MCP"> </a>

🤖 Technologies Used

• —Orchestration: FastMCP + Claude Desktop
• —Persona Generation: OpenAI GPT-4o (via API)
• —Ad Image Generation: Google Gemini Nano Banana / Imagen (AWS)
• —Backend: FastAPI + Python

⚡ AI & Infrastructure

• —Reasoning: Anthropic Claude 4.5 Sonnet
• —Image Generation: Sipify Proprietary Engine (AWS GPU)
• —Storage: AWS S3 & CloudFront
• —Scraping: Custom Python Scraper

🪄 The Magic: Brand URL -> Full Campaign

Most AI tools need complex prompts. Sipify just needs a URL.

We built a multi-step Agentic Workflow that thinks like a human Creative Director:

1. 1.👀 The Eye (Scraping & Vision):
2. 2.The agent autonomously visits the brand's URL (e.g., lego.com).
3. 3.It extracts the soul of the brand: Logo, Color Palette (Hex codes), and Visual Style.

1. 1.🧠 The Brain (Strategy & Copy - OpenAI):
2. 2.Using the scraped data, it devises a "Marketing Persona."
3. 3.Example: For Lego, it didn't just say "buy blocks." It created a "Master Builder" persona focused on creativity, energy, and vibrant play.

1. 1.🎨 The Artist (Visual Production - Gemini):
2. 2.It translates the Persona into high-fidelity image prompts.
3. 3.It autonomously generates infinite assets in parallel—tailored to any dimension (Square, Story, Landscape) based on real-time requirements

Result: A generic request ("make ads for lego") turns into a production-ready campaign in 60 seconds.

🗂️ Project Structure

sipify-mcp/
├── server.py # Main MCP Server 
├── README.md # Documentation & Demo
├── requirements.txt #Requirements

Working FlowChart-

[image]

💻 Installation (Run Locally)

1. 1.Install Dependencies

    pip install fastmcp httpx

1. 1.Run the Server

    python server.py

1. 1.Connect to Claude Add this to your claude_desktop_config.json:

    {
      "mcpServers": {
        "sipify": {
          "command": "python",
          "args": ["/path/to/server.py"]
        }
      }
    }

🏆 Social & Submission

• —Social Post: https://www.linkedin.com/posts/tendsxgauravmcphackathon-aiagents-anthropic-ugcPost-7400875610627309568-ucGS?utmsource=share&utmmedium=memberdesktop&rcm=ACoAAFCXa34BAU54TiYthI7ZC1ZQ2kVPbZYBuxs
• —Team: @inGaurav
• —Track: Building MCP (building-mcp-track-01) (building-mcp-track-creative)

Note: This project uses a live connection to Sipify's AWS backend for demonstration purposes.