What problem does it solve? When a bug report or feature request arrives, it is often unclear which layer of the system is at fault or what kind of test would actually catch the regression. This Skill turns ambiguous reports into a bounded, duplicate-checked regression plan instead of a pile of ad-hoc test tasks. ## Core Features & Use Cases - Source-of-truth classification: Traces a symptom inward through UI, live state, sync, imports, API contract, auth, and operational layers to find the first layer that can produce the wrong result. - Duplicate detection: Searches existing project tasks, tests, and specialist skills before proposing new work, so no duplicate regression tasks are created. - Smallest effective test selection: Recommends exactly one primary test type (unit, integration, browser, real-mobile, visual, accessibility, or manual) with justification, and routes to specialist skills like sync-invariant-check or import-bug-investigation. - Use Case: A report says a deleted recipe reappeared after another device synced. The Skill classifies it as a sync issue, routes to sync-invariant-check, and recommends a two-client integration test plus a data-heal check if production data was already resurrected. ## Quick Start Ask the assistant to triage a bug report into a regression plan, for example by describing the observed behavior, expected behavior, and where it occurs.
