An AI-powered visual dashboard to explore, document, and audit your DBT (Data Build Tool) projects using OpenAI or Azure OpenAI's GPT models.
Built with Streamlit, this tool provides natural-language summaries, lineage graphs, column-level usage, and model metrics — all generated from your DBT project structure and SQL.
- ✨ Features
- 🚀 Live Demo
- 🚀 Demo Preview
- 📂 Project Structure
- ⚙️ Setup Instructions
- 🧠 How GPT is Used
- 📁 Supported Input Types
- 💡 Use Cases
- 📤 Export Capabilities
- 🛠 Tech Stack
- 🪪 License
- 🙌 Acknowledgements
| Feature Category | Description |
|---|---|
| 📄 GPT Model Summarizer | Generate GPT-based explanations of each DBT model |
| 📊 Lineage Graph | Interactive visual DAG of ref() relationships |
| 🌳 Ref Tree Explorer | Explore upstream/downstream dependencies per model |
| 📁 Folder Summary | GPT summary of entire folders (e.g. staging, marts, reporting) |
| 📋 Model Metadata Table | View description, columns, references for all models |
| 🔍 Search Tool | Search by column name, model name, or SQL content |
| 🧬 Column Lineage Viewer | Trace how each column is used across your DBT project |
| 🧭 Categorization Viewer | Categorize models as staging, dim, fact, or other based on prefix |
| 📈 Complexity Analyzer | Auto-classify models based on size and reference depth |
| 📤 Export Options | Export GPT summaries, lineage graphs, and metrics |
dbt-model-analyzer/
├── app.py # Main Streamlit app
├── config.json # LLM config file
├── requirements.txt # Python dependencies
├── screenshots/ # UI screenshots (optional)
└── README.md| Input Type | Description |
|---|---|
| 📁 ZIP File Upload | Upload your DBT project as a .zip archive |
| 🌐 GitHub Clone | Clone any public GitHub repo containing DBT code |
- 🧠 Understand unfamiliar DBT models and folders
- 🗂 Auto-document large legacy DBT projects
- 🔍 Trace where each column is used
- 📈 Find complex models with too many refs or lines
- 📊 Export summaries for reporting or onboarding
- 📁 Explore staging vs mart models folder by folder
| Type | Format |
|---|---|
| GPT Summaries | .md |
| Lineage Graph | .html |
| Model Metadata | .csv or table |
| Column Lineage | .csv or table |
| Complexity Metrics | DataFrame view |
- Streamlit – frontend dashboard framework
- OpenAI Python SDK – GPT-3.5/4 integrations
- PyVis – interactive visual DAGs and networks
- NetworkX – lineage and DAG graph construction
- YAML / Pandas / Regex – model and metadata parsing