Context-Aware User Error Messaging

Many applications fail in ways that are invisible or misleading to the people using them. This is especially true for software deployed directly at the end user—running on their device or in their local environment—where failures may originate in a third-party API, an external service, or a connected peripheral device, yet the user only sees a vague message like “something went wrong.” This project shall explore how runtime failures in such deployed applications can be translated into clear, human-oriented feedback that reflects the full context in which the application operates—not just the app itself, but the integrations and devices it depends on. The improved error analysis and messaging shall run offline as part of the deployed application itself, without relying on a network connection to interpret or explain what went wrong. The goal is to help end users understand what actually went wrong and what they can reasonably do next. By drawing on knowledge of the application together with documentation and behaviour of external APIs and peripheral hardware, the system aims to explain failures that may surface far from their true cause or repeat as familiar integration patterns. For the hackathon, the focus is on investigating whether richer contextual understanding—available locally at the point of failure—can turn cryptic runtime errors into meaningful, actionable user experiences rather than dead ends.