SKaush/lyra-agentic-ai-framework

Agentic Data Testing Platform

An AI-assisted automated testing framework that transforms SQL mapping documents and raw source data into executable test scenarios, then validates them against Snowflake target tables.

Features

• —Mapping Extraction - Parse Excel mapping documents to extract source tables, JOINs, WHERE filters, and column transformations
• —Data Quality Profiling - Profile raw CSV data for nullness, cardinality, column health, and generate variance samples
• —Scenario Generation - Automatically build INSERT, UPDATE, and DELETE test scenarios from joined/profiled data
• —Snowflake Validation - Execute scenarios against live Snowflake target tables and compare expected vs actual results
• —Coverage Reporting - Measure how much of the mapping logic (columns, joins, filters, transformations) is exercised by test scenarios
• —Failed Scenario Diagnostics - Detailed root cause analysis for failed validations with suggested fixes

Project Structure

.
├── app.py                    # Streamlit UI entry point
├── core/
│   ├── profiling/            # Data quality & coverage
│   ├── snowflake/            # Database operations
│   ├── scenarios/            # Test scenario generation
│   ├── agent.py              # OpenAI integration
│   └── audit.py              # Audit logging
├── cli/
│   └── run_agent.py          # CLI entry point
├── data/
│   ├── sample/               # Demo data
│   └── mapping_document/     # Excel mapping files
└── requirements.txt

Quick Start

Streamlit UI

pip install -r requirements.txt
streamlit run app.py

CLI

python cli/run_agent.py

Environment Variables

Copy .env.example to .env and configure:

HuggingFace Spaces Deployment

This app is configured for HuggingFace Spaces (Streamlit SDK). Configure the environment variables above as Secrets in the Space settings.

Tech Stack

Python | Streamlit | Snowflake | OpenAI | pandas | openpyxl