https://insightapp-sreeyasrikanth.streamlit.app/
InsightQA is a QA automation assistant that converts product documents into structured test cases and executable Selenium-style scripts. It uses Retrieval-Augmented Generation and a Llama3 LLM backend to generate grounded, document-based outputs.
Users can upload:
- A primary file (HTML UI, API spec, flow, etc.)
- Multiple supporting documents (PDF, TXT, MD, JSON, etc.)
Each upload creates a separate Knowledge Base (KB).
The system:
- Extracts and parses text
- Chunks and embeds documents using Sentence Transformers
- Stores vectors in ChromaDB
- Maintains metadata for all uploaded files
Each KB can be viewed, renamed, or deleted.
Given a natural-language request such as:
“discount code validation”
InsightQA retrieves the most relevant context and generates structured JSON test cases including:
- Test ID
- Test Scenario
- Preconditions
- Steps
- Expected Result
- Context references
All test cases are grounded in the uploaded documents.
InsightQA can convert a selected test case into a runnable Python script.
Locators and actions are inferred from the uploaded HTML file.
- Upload main and support files
- Build a Knowledge Base
- Enter a feature to test
- Retrieve relevant context
- Generate structured test cases
- Select a test case
- Generate the corresponding script
- Streamlit
- FastAPI
- ChromaDB
- SQLite
- Sentence-Transformers
- Llama3
MIT License