root-cause-tracing

Root-cause tracing follows a five-step workflow: Observe the symptom, identify the Immediate failure point, Trace Up through the call chain, Keep Going backward through layers, and Find Source at the original trigger. This backward tracing ensures fixes target the actual bug origin, not surface errors, especially across modular codebases and asynchronous flows.

Symptom-based fixes address visible errors without locating their origin, leading to recurring bugs. Root-cause tracing identifies the original trigger in the call chain, ensuring permanent fixes. The five-step workflow prevents symptom-focused edits by enforcing systematic backward investigation through all code layers.

Yes. Root-cause tracing is designed for codebases with modular components and asynchronous data flows. The systematic five-step workflow traces backward through these complex architectures to locate the original trigger, making it effective for distributed logic and event-driven systems.

Start by noting the symptom, then invoke mcp__zen__debug to structure your investigation with hypothesis-driven evidence gathering. Follow the five-step trace workflow—Observe, Immediate, Trace Up, Keep Going, Find Source—to systematically work backward through the call chain until you locate the root cause.

The first visible error is often a downstream consequence, not the source. Backward tracing through the call chain reveals where the bug actually originated, preventing temporary fixes that mask deeper issues. This approach ensures changes address the root trigger and eliminate the bug permanently across all its manifestations.

Root-cause tracing integrates with mcp__zen__debug for hypothesis-driven investigation and structured evidence gathering. The five-step workflow complements standard debugging by enforcing systematic backward tracing through call chains, rather than replacing your tools—it layers systematic methodology on top of existing debugging.