AI-powered software testing assistant: a full-cycle QA assistant on GitHub

⚡ The gist in 5 seconds - AI-powered software testing assistant has been published on GitHub — a full-stack application for QA engineers: from requirements analysis to bug diagnostics. - The project is available in the sirivennelaa2004-create repository, runs locally (Python 3.10+, Node.js 18+) or via Docker Compose. - By default it runs the free MockAIProvider — a provider API key is required for real AI. ### 🔍 What was found The project is described as a production-grade full-stack web application aimed at software QA engineers and testers. Its core is a pipeline: user requirements are converted into structured test scenarios and test cases, then test runs are executed, defects are logged, and AI is used for requirements analysis and root cause diagnostics of bugs. The stack is serious. Backend: Python, FastAPI, SQLAlchemy 2.0, Pydantic v2, authentication via PyJWT, password hashing with Passlib (Bcrypt), SQLite/PostgreSQL databases. Frontend: React 18, Vite, React Router v6, Axios, Lucide Icons, and a custom CSS system with a dark theme and glassmorphism. The AI subsystem is built as a pluggable AIService facade: OpenAI, Gemini, Ollama, and Groq are supported, plus a free MockAIProvider as a fallback; there is a resilient JSON parser with response extraction and "repair". Tests: a pytest suite with 100% pass rate covering auth, authorization, CRUD, the AI parser, execution, and the bug workflow. The workflow looks like this: requirement → AI analysis → scenario generation (positive, negative, edge, boundary, security) → test cases with TC-ID, objective, preconditions, steps, and expected result → execution (Passed/Failed/Blocked) → bug reports (auto-created on failure or manually) → AI diagnostics with root cause, suggested Severity and Priority, and fix steps → quality dashboard with pass rate, status distribution, and activity logs. ### 💡 Why it matters The tool covers the entire QA cycle in a single application: instead of a set of disparate utilities, the developer gets a unified pipeline from requirement to dashboard. What's practically valuable is that the AI layer is pluggable — you can choose OpenAI, Gemini, Ollama, or Groq, and work on MockAIProvider without keys. Docker Compose brings up PostgreSQL and the backend with a single command, Swagger UI is available at /docs, health check at /health, and the frontend at localhost:5173. ### 🧩 Context The project was published by user sirivennelaa2004 (repository sirivennelaa2004-create/ai-powered-software-testing-assistant). Configuration via.env: DATABASE_URL, SECRET_KEY, AI_PROVIDER (mock by default), AI_API_KEY, AI_MODEL_NAME (gpt-4o-mini), AI_BASE_URL (https://api.openai.com/v1). How well the tool handles real-world tasks cannot be judged from the repository description alone.

🔗 Read on github.com