The '0 tests passed is not a pass' check is the one we hit hardest in Electron work: an AI-generated IPC handler looks correct, but it never validates the renderer payload, and the integration test that should catch that is mocked away. The habit of asking 'which command ran and what was the exit code' generalizes well to desktop — especially around auto-update and crash-reporting code, where the happy path is easy to generate and the failure path (network down, partial download, stale dump) is exactly what gets skipped. Pairing each AI change with a forced failure-path test has saved us more than the happy-path tests did.
We're a place where coders share, stay up-to-date and grow their careers.
The '0 tests passed is not a pass' check is the one we hit hardest in Electron work: an AI-generated IPC handler looks correct, but it never validates the renderer payload, and the integration test that should catch that is mocked away. The habit of asking 'which command ran and what was the exit code' generalizes well to desktop — especially around auto-update and crash-reporting code, where the happy path is easy to generate and the failure path (network down, partial download, stale dump) is exactly what gets skipped. Pairing each AI change with a forced failure-path test has saved us more than the happy-path tests did.