Claudehopper

🚀 ClaudeHopper - AI-Powered Construction Document Assistant

ClaudeHopper is a specialized Model Context Protocol (MCP) server that enables Claude and other large language models to directly interact with construction drawings, documents, and specifications. It achieves efficient document querying through advanced RAG (Retrieval-Augmented Generation) and hybrid search.

✨ Features

• 🔍 Vector-based search optimizes the retrieval of construction documents for CAD drawings, plans, and specifications.
• 🖼️ Visual search function can find similar drawings based on text descriptions.
• 🏢 Metadata extraction specifically tailored to construction industry document formats.
• 📊 Efficient token usage through intelligent document chunking and classification.
• 🔒 Security is ensured through local document storage and processing.
• 📈 Supports multiple drawing types and construction disciplines (structural, civil, architectural, etc.).

🚀 Quick Start

Prerequisites

• Node.js 18+
• Ollama for local AI models
  • Required models: nomic-embed-text, phi4, clip
• Claude Desktop App
• For image extraction, install Poppler Utils (pdfimages command).

One-Click Deployment

1. Download ClaudeHopper.
2. Run the installation script:

cd ~/Desktop/claudehopper
chmod +x run_now_preserve.sh
./run_now_preserve.sh

This script will:

• Create the necessary directory structure.
• Install the required AI models.
• Process your construction drawings.
• Configure the Claude Desktop App to use ClaudeHopper.

Add Documents

Place your construction drawings in the following folders:

• Drawings: ~/Desktop/PDFdrawings-MCP/InputDocs/Drawings/
• Specifications: ~/Desktop/PDFdrawings-MCP/InputDocs/TextDocs/ After adding the documents, run:

./process_pdfdrawings.sh

💻 Usage Examples

Using ClaudeHopper with Claude

Try the following sample questions in the Claude Desktop App:

“What construction drawings do we have?”
“Show the structural details of the foundation system.”
“Find the drawings showing the concrete foundation and its dimensions.”
“Search for the layout plan of the water supply and drainage station.”
“What are the specifications for the building paint?”
“Find all content discussing the fire protection system.”

🔧 Technical Details

ClaudeHopper uses a multi-stage pipeline to process construction documents:

1. Document Analysis: Analyze the structure and content types of PDF documents.
2. Metadata Extraction: AI-assisted extraction of project information, drawing types, and disciplines.
3. Content Chunking: Intelligent document chunking and classification.
4. Image Preprocessing: Optimize image quality for AI model processing.
5. Model Inference: Use Claude for context understanding.

📋 Testing Image Search

To test the image search function, run the following command:

python test_image_search.py

Ensure that the necessary dependencies are installed and follow the instructions in the script.

📄 License

This project is licensed under the MIT License.