What problem does it solve? When facing bugs, test failures, or regressions, developers often jump straight to editing code and hope the symptom disappears, which hides the root cause and lets it resurface elsewhere. This Skill forces a disciplined sequence: reproduce the issue reliably, locate the exact file and line, form a falsifiable root-cause hypothesis, and only then write a fix with a regression test. ## Core Features & Use Cases - Four gated phases: Reproduction, localization, hypothesis with falsification, and fix — each phase has a required output and you cannot skip ahead. - Root-cause discipline: Prohibits code changes before the cause is identified, and requires stating the cause in the form "because <code> under <condition> does <behavior>, we observe <symptom>". - Pitfall detection: Flags common anti-patterns like fixing symptoms instead of causes, trusting stale documentation, changing multiple variables at once, and believing green tests prove correctness. - Use Case: A test suite suddenly fails after a refactor. Instead of tweaking code until it passes, you first reproduce the failure with a stable command, bisect to the offending line, write a failing test that captures the regression, then fix and verify the full suite passes. ## Quick Start Use the debug-systematically skill to investigate why my login endpoint returns a 500 error only under concurrent requests.
