root-cause-tracing

Root-cause tracing guides you backward from the error symptom through the call chain to find the original trigger. Instead of fixing the error where it appears, you identify where invalid data or incorrect behavior originated, making the fix structural and permanent.

Use backward tracing when errors occur deep in execution across long call stacks, especially in tests where data corruption's origin is unclear. Tracing to the source prevents the same bug from surfacing elsewhere, whereas defensive checks only mask symptoms.

Add diagnostic logging at key decision points and capture full stack traces when errors occur. This instrumentation creates breadcrumbs through the execution path, revealing where state changed unexpectedly and narrowing the search space for the original trigger.

Use bisection techniques to systematically narrow which test or code pollutes the state. The Skill references a polluter-finding script that isolates the exact source, letting you fix the root cause rather than working around side effects.

Yes. By tracing backward through the call stack, you map how different execution paths converge at the error point, revealing which path introduced the invalid state. This uncovers the common origin even when symptoms vary.