AI Instructions for CAD2DATA Repository
Overview
This folder contains instructions for AI coding assistants (Claude Code, Opencode, Google Antigravity, Cursor, Copilot, Windsurf, Aider, Cline, and others) to effectively work with the CAD2DATA pipeline repository.
Purpose
All tools in this repository are open-source and designed to be:
- Freely integrated into any solution
- Extended and customized by AI assistants
- Used as building blocks for construction automation workflows
- Called directly from CLI by AI assistants
Repository Philosophy
This repository follows the Data-Driven Construction approach as described in the book DATA_DRIVEN_CONSTRUCTION_BOOK.txt located in this folder. The book serves as a guiding beacon for automation processes in construction companies.
Key Principles
- Data as Foundation: Construction data is the new oil - extract, transform, and utilize it effectively
- Open Formats: Convert proprietary CAD/BIM to open formats (XLSX, DAE, IFC, CSV, HTML, PDF)
- No Vendor Lock-in: Works offline without Autodesk licenses or cloud APIs
- Automation First: Every repetitive task should be automated
- AI-Ready Pipelines: Design workflows that can be enhanced by AI assistants
Repository Structure
cad2data-Revit-IFC-DWG-DGN-pipeline/
├── AI_AGENTS_INSTRUCTIONS/ # Instructions for AI assistants
├── DDC_Converters_Windows_Packages/ # Windows converters (.exe)
│ ├── DDC_CONVERTER_DGN/ # DGN converter (DgnExporter.exe)
│ ├── DDC_CONVERTER_DWG/ # DWG converter (DwgExporter.exe)
│ ├── DDC_CONVERTER_IFC/ # IFC converter (IfcExporter.exe)
│ ├── DDC_CONVERTER_REVIT/ # Revit converter (RvtExporter.exe)
│ ├── DDC_CONVERTER_Revit2IFC/ # Revit to IFC (RVT2IFCconverter.exe)
│ └── DDC_Update_Revit_from_Excel/ # Excel to Revit sync
├── DDC_Converters_Linux_Packages/ # Linux converters (.deb)
│ ├── ddc-qdrant/ # Qdrant vector DB package
│ ├── ddc-cwicr-cli/ # Semantic search CLI
│ └── ddc-cwicr-data/ # Language data packages (9 languages)
├── DDC_n8n_Workflows&Pipelines/ # n8n automation workflows (9 workflows)
├── DDC_Python_pipelines/ # Python pipeline examples
├── DDC_in_additon/ # Additional utilities
└── Sample_Projects/ # Test data
Supported Input/Output Formats
Input Formats
| Format | Versions | Converter |
|---|
| Revit (.rvt, .rfa) | 2015-2026 | RvtExporter.exe, RVT2IFCconverter.exe |
| IFC (.ifc) | IFC2x3, IFC4, IFC4x1, IFC4x3 | IfcExporter.exe |
| AutoCAD (.dwg) | 1983-2026 | DwgExporter.exe |
| MicroStation (.dgn) | V7, V8 | DgnExporter.exe |
Output Formats
| Format | Description | Produced By |
|---|
| XLSX | Excel database (elements as rows, properties as columns) | All converters |
| DAE | Collada 3D geometry with element IDs matching XLSX | RvtExporter, IfcExporter |
| PDF | Drawings and sheets | RvtExporter, DwgExporter |
| IFC | IFC2x3, IFC4, IFC4.3, IFCXML, IFCZIP, HDF5 | RVT2IFCconverter |
| HTML | Interactive reports with charts | n8n workflows |
| CSV | Simple tabular data | Post-processing |
Available Instructions
- INSTRUCTIONS.md - This file (general overview)
- TOOLS_OVERVIEW.md - Detailed tool descriptions and CLI syntax
- CLAUDE.md - Claude Code specific instructions
- OPENCODE.md - Opencode specific instructions
- ANTIGRAVITY.md - Google Antigravity specific instructions
- DATA_DRIVEN_CONSTRUCTION_BOOK.txt - The guiding book
For AI Assistants
When working with this repository:
- Read the book -
DATA_DRIVEN_CONSTRUCTION_BOOK.txt provides context for construction automation
- Use CLI tools directly - All converters have command-line interfaces that AI can execute
- Respect open formats - Output to XLSX, DAE, IFC, HTML, PDF
- Think data-first - Extract structured data (DataFrames), not just geometry
- Automate workflows - Use n8n JSON templates as starting points
Quick CLI Examples
Windows (.exe)
# Convert Revit to XLSX + DAE
RvtExporter.exe "C:\Projects\Building.rvt"
# Convert Revit with all options
RvtExporter.exe "C:\Projects\Building.rvt" complete bbox schedule sheets2pdf
# Convert Revit to IFC
RVT2IFCconverter.exe "C:\Projects\Building.rvt"
# Convert IFC to XLSX + DAE
IfcExporter.exe "C:\Projects\Model.ifc"
# Convert DWG to XLSX + PDF
DwgExporter.exe "C:\Projects\Plan.dwg"
# Convert DGN to XLSX
DgnExporter.exe "C:\Projects\Bridge.dgn"
Linux (.deb packages)
Install from APT repository:
# Add DDC APT repository
echo "deb [trusted=yes] https://pkg.datadrivenconstruction.io stable main" \
| sudo tee /etc/apt/sources.list.d/ddc.list
sudo apt update
# Install converters (SDK dependencies are installed automatically)
sudo apt install ddc-rvtconverter ddc-dwgconverter ddc-ifcconverter \
ddc-dgnconverter ddc-rvt2ifcconverter
Run converters:
# Revit to XLSX + DAE
ddc-rvtconverter input.rvt output_dir/
# DWG to XLSX
ddc-dwgconverter input.dwg output_dir/
# IFC to XLSX + DAE
ddc-ifcconverter input.ifc output_dir/
# DGN to XLSX + DAE
ddc-dgnconverter input.dgn output_dir/
# Revit to IFC
ddc-rvt2ifcconverter input.rvt output_dir/
System requirements: Ubuntu 20.04+ / Debian 11+ (amd64)
Linux — DDC CWICR Semantic Search
# Install Qdrant vector database
sudo apt install ddc-qdrant
# Install language data (each ~1 GB of pre-computed embeddings)
sudo apt install ddc-cwicr-en # English (CAD, Toronto)
sudo apt install ddc-cwicr-de # German (EUR, Berlin)
sudo apt install ddc-cwicr-ru # Russian (RUB, St. Petersburg)
# Install search CLI
sudo apt install ddc-cwicr-cli
# Search (keyword — no API key needed)
ddc-search --keyword "concrete foundation"
# Search (semantic — requires OpenAI API key)
export OPENAI_API_KEY=sk-...
ddc-search "reinforced concrete foundation"
# JSON output
ddc-search --json --limit 10 "floor tiles"
Available CWICR languages: en, de, ru, fr, es, ar, zh, pt, hi (55,719 work items per language)
n8n Workflows - Visual Process Templates
The n8n_*.json files are visual representations of process logic - not the final solution, but a foundation that can be implemented in any language or application.
Use these workflows as:
- Reference architecture for building your own pipelines
- Visual documentation of data processing logic
- Starting point for Python, JavaScript, C#, or any other implementation
- Templates for integration into enterprise systems
Available Process Templates
- Basic Conversion - Simple file conversion logic
- Advanced Settings - Configurable export options
- Batch Processing - Multiple files with reporting
- Validation - BIM data quality checks
- AI Classification - LLM + RAG element classification
- Cost Estimation - Price calculation with DDC CWICR database
- Carbon Footprint - CO2 emissions calculation
- ETL for LLM - Data preparation for AI tasks
- QTO Reports - Quantity take-off report generation
The logic shown in n8n can be reproduced in: Python, JavaScript/Node.js, C#, Go, Rust, or any language that can call CLI tools and process XLSX/JSON.
License
All tools are open-source (MIT). See LICENSE file for details.
"If data is the new oil, we need to learn to define it, find it, mine it, refine it, to make it valuable."
— Ralph Montague