ct-systematic-debugging

Follow the four phases: first investigate the root cause by reading errors fully, reproducing consistently, and checking recent changes. Then analyze patterns, form one hypothesis, test it minimally, and only then implement a fix with a failing test first.

Still complete root cause investigation first. The skill states that systematic debugging is faster than guess-and-check thrashing, and symptom fixes are classified as failure regardless of urgency or pressure from others.

Use the included find-polluter.sh bisection script with the polluted path and a test pattern. It runs test files one by one and stops at the first test that creates the unwanted file or directory.

Replace setTimeout or sleep calls with condition-based waiting that polls for the actual condition you care about, such as an event appearing or state changing. The reference docs include waitForEvent and waitForEventCount implementations.

Stop attempting more fixes and question the architecture. Three or more failures indicate a structural problem, so discuss whether the pattern is fundamentally sound with your team before trying another fix.

Never, according to the skill. Simple bugs have root causes too, emergencies make guessing more costly, and skipping investigation guarantees rework. The process is designed to be fast for simple issues.