What problem does it solve? Failing Playwright tests are often diagnosed from the error message alone, producing wrong root-cause classifications — real app bugs get hidden by test edits, and test issues get misreported as app bugs. This Skill enforces a structured, evidence-first diagnostic pipeline so every failure is classified from observed page state rather than guesswork. ## Core Features & Use Cases - Dual entrypoints: Handles both local failures (artifacts already on disk) and CI pipeline failures (artifacts pulled down via the GitHub CLI, pinned to the run's commit and dependency versions). - Mandatory evidence floor: Requires inspection of the trace, the UI/DOM at failure, and the browser console before any root cause is proposed, with explicit handling for missing traces and expired artifacts. - Classification and healing: Groups failures by root cause, classifies them as test issue, app bug, framework defect, or ambiguous, then applies the matching heal strategy or files an evidence-backed bug report. - Use Case: When a nightly CI regression goes red, dispatch a subagent with this Skill to download the run's artifacts, inspect the failing attempt's trace and screenshots, and report whether the failures are stale selectors, a genuine app regression, or a framework version mismatch. ## Quick Start Dispatch a subagent with the failure-diagnosis skill to investigate why the nightly CI regression run failed and report the root cause with evidence.
