WebSpector is an innovative, agentic AI-powered Quality Assurance (QA) platform designed to automate and elevate the process of website testing. Traditional software testing relies heavily on manually written, brittle scripts that require constant maintenance. WebSpector eliminates this bottleneck by deploying autonomous AI agents capable of visually and structurally understanding any web application. Operating through a sophisticated four-stage pipeline Understand, Plan, Execute, and Analyze the system navigates websites, captures UI states across multiple viewports (Desktop, Tablet, Mobile), and utilizes Large Language Models (Google Gemini and Anthropic Claude) to dynamically generate and execute test plans. It rigorously checks for accessibility flaws, responsive design issues, form validation errors, broken links, and JavaScript exceptions without human intervention. Finally, the platform synthesizes its findings into a comprehensive, deduplicated, and categorized bug report with actionable fix suggestions. By combining browser automation with deep-reasoning LLMs, WebSpector acts as a tireless, expert QA engineer, significantly reducing testing time and improving software reliability for modern web development.
Tools: Python, FastAPI, React, Playwright, Supabase, PostgreSQL, Docker, Large Language Models
Department: Department of Computer Science
Poster
