Mock interview feedback: 'your design doesn't handle edge cases' — how to improve?

I recently got feedback from a mock interview that my design 'doesn't handle edge cases' well. My solutions often work for the happy path, but I struggle to systematically think through all the failure modes and less common scenarios like node failures mid-write, clock skew issues in distributed systems, or cache-DB consistency problems. It feels like there are so many things to consider. Does anyone have a mental checklist or a structured approach for identifying and addressing edge cases during a system design? How do you ensure you're not just thinking about the ideal scenario? I want to move beyond just mentioning 'error handling' and actually integrate robust solutions for these trickier situations.