Some technical details - Under the hood, JAIPilot uses IntelliJ’s PSI for code context extraction and wraps Gemini in an agentic retry loop until the generated

Some technical details - Under the hood, JAIPilot uses IntelliJ’s PSI for code context extraction and wraps Gemini in an agentic retry loop until the generated tests compile. Code here : https://github.com/skrcode/java-auto-unit-tests https://github.com/skrcode/java-auto-unit-tests Would be very interested to hear if anyone has tried different approaches for reliable AI-generated tests.