Compare e2e captured translations against the baoyu-translate skill ground truth (tests/reference/article-zh-CN-<mode>/translation.md) for all three translation modes, computing both cheap metrics (character jaccard, Chinese 2-gram overlap, paragraph drift) and Claude judge scores across 5 dimensions (accuracy, terminology consistency, cultural annotation, fluency, faithfulness). Writes tests/e2e-out/<ts-sha>/quality.md. Use only after /e2e-diagnostician returns verdict=PASS — when the gate fails, this skill must not run. Triggers on /e2e-quality, "run quality", "compare translations", "judge e2e quality", or after a successful /e2e-diagnostician run.
Aggregate e2e-diagnostician's diagnosis.md and e2e-quality's quality.md into a final acceptance-form e2e-report.md at tests/e2e-out/<ts-sha>/e2e-report.md. Use only when the full pipeline has run (runner → diagnostician PASS → quality); when the gate fails, e2e-diagnostician already wrote e2e-report.md and this skill must not overwrite it. Triggers on /e2e-reporter, "generate e2e report", "summarize e2e results", or after a successful /e2e-quality run.
Diagnose a Baoyu Translate e2e-runner artifact set and act as the gate that decides whether downstream quality + reporter agents run. Reads tests/e2e-out/<ts-sha>/ runner products (captured.md / status.md / console.jsonl / failure.md?), produces a diagnosis.md, and — when the gate fails — additionally writes an e2e-report.md in acceptance form so the upstream caller can stop the pipeline. Use when the user asks to "diagnose e2e", "run e2e gate", "为什么 e2e 没过", or after a /e2e-runner run is complete.
Reproduce the real user path of the Baoyu Translate Chrome extension via chrome-devtools-mcp, run the three translation modes (quick / normal / refined) in isolation, and persist every artifact to tests/e2e-out/<iso8601-timestamp>_<git-sha>/<mode>/. Use when the user asks to run an e2e, debug a translation regression, refresh the ground-truth-style captured translations, or wants to verify the extension on a real page. Triggers on /e2e-runner, "run e2e", "end-to-end test the extension", "capture the live translation", or "refresh e2e-out".