Identifies the root cause of failing tests using a structured 4-phase semi-formal reasoning process.
The locate-fault skill transforms Claude into a systematic debugging expert by enforcing a rigorous approach to fault localization. Instead of stopping at the immediate crash site, it guides the AI through test semantics analysis, hypothesis-driven code path tracing, and divergence analysis. This method ensures that complex logic errors and indirection bugs—where the failure occurs far from the actual mistake—are traced back to their source, providing developers with evidence-based ranked predictions and a clear path to a permanent fix.
Key Features
01Hypothesis-driven file exploration to prevent aimless code reading
02Ranked bug predictions with explicit confidence levels and supporting claims
