What problem does it solve? When a CI run goes red, engineers waste time digging through logs and artifacts to figure out whether the failure is a real application bug, a flaky test, or an infrastructure issue. This Skill automates that investigation and delivers a structured diagnosis directly on the pull request. ## Core Features & Use Cases - Evidence Collection: Pulls failed run logs and the playwright-report artifact via the gh CLI or GitHub MCP, extracting the failing test, expected vs actual values, and trace paths. - Cross-Referenced Classification: Compares the spec, page object model, feature acceptance criteria, and repo conventions to classify the failure as an app bug, test issue, or CI/infra problem with a source-level root cause. - Structured PR Reporting: Posts a templated comment with run ID, classification, root cause, suggested fix, and evidence, then routes confirmed app bugs to a Jira bug-reporting workflow. - Use Case: A pull request's E2E check fails overnight. The Skill downloads the report artifact, identifies that a locator no longer matches after a UI change, classifies it as a test issue, and proposes a minimal patch for human review. ## Quick Start Triage the failing E2E run on this pull request and post a diagnosis comment with the root cause and suggested fix.
