browser-e2e-test
Build kokoro-js-jp and run its real-browser e2e suite (Worker + WASM openjtalkjs g2p + ONNX kokoro-js synthesis, English and Japanese). Use when asked to verify this package actually works in a browser, to check whether a change broke synthesis, or to add browser-level regression coverage for a bug in the g2p/synthesis pipeline that vitest's pure-function tests can't catch. Not for typecheck/lint/unit-test requests — use npm run typecheck / npm test for those.
中文简介正在准备中。当前显示原文说明。
包含的文件（1）
- SKILL.md1.9 KB
SKILL.md（原文）
安装前可以查看交给 AI 代理的指令内容。
Runs kokoro-js-jp's browser e2e suite (test/e2e/tts.spec.ts, Playwright) and reports the
result. This is the only verification path that exercises the actual Worker + WASM +
ONNX pipeline — npm test (vitest) only covers pure functions. See AGENTS.md at the repo
root for the full rationale (this pipeline had a long-standing bug that silently broke
every KokoroJP.load() call and only real-browser testing could have caught it).
Steps
- Confirm you're in the
kokoro-js-jprepo (package.jsonname field). If not, find it — it's a separate repo from whatever else is open, commonly at~/public_html/kokoro-js-jp. - Delegate to the
browser-e2e-testersubagent (.claude/agents/browser-e2e-tester.md) to build, run, and — if something fails — investigate and either fix it or report exactly what's broken and where. Don't duplicate its debugging logic here; that file is the source of truth for how to interpret failures. - Relay the agent's result plainly: pass/fail, and for a fail, what broke and what (if anything) was fixed. Don't editorialize beyond that.
Args
If the user names a specific scenario to check (e.g. "does the French voiceId error case
still work", "check the dictionary reachability regression"), pass that through to the
subagent as extra context rather than just re-running the whole suite blind — it can target
npx playwright test -g "<pattern>" if that's faster for the ask.
评价
还没有评价。欢迎分享你的使用感受。
评论
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。