Cursor Debugging Prompt Pack

Keep AI debugging anchored to a reproducible failure and a falsifiable hypothesis.

What the resource contains

10 debugging prompts

Hypothesis log

Frontend and API variants

Regression-test handoff

Selected preview

Minimal reproduction

Reduce the symptom before proposing a fix.

Turn the report into: expected, actual, minimal steps, first known bad boundary, and one observable assertion. Inspect relevant logs and code. If the failure cannot be reproduced, list the missing evidence and add targeted instrumentation—do not guess a fix.

Trace the value

Follow one wrong value through the system.

Choose the smallest incorrect output. Trace where it is created, transformed, cached, serialized, and rendered. At every boundary record the expected and observed value. Stop at the earliest divergence and test the suspected invariant.