Agent Skills
› AjayIrkal23/agentic-mercy-10x
› webapp-testing
webapp-testing
GitHub用于在真实浏览器中端到端测试Web应用，通过Playwright驱动流程、验证UI并捕获截图、控制台及网络证据。负责本地服务器生命周期管理，需结合调试和代码执行技能使用。
触发场景
复现前端问题
使用Playwright验证本地UI流程
捕获截图、控制台输出或DOM状态
安装
npx skills add AjayIrkal23/agentic-mercy-10x --skill webapp-testing -g -y
SKILL.md
Frontmatter
{
"name": "webapp-testing",
"schema": 1,
"category": "testing",
"surfaces": [
"backend",
"frontend"
],
"triggers": {
"paths": [
".spec.",
".test.",
"__tests__",
"_test.ts",
"_test.tsx"
],
"intents": [
"testing"
],
"keywords": [
"analyze",
"anything",
"automation",
"behavior",
"browser",
"browsers",
"building",
"capture",
"capturing",
"chrome",
"console",
"debugging",
"devtools",
"dom",
"errors",
"inspect",
"inspecting",
"local",
"mcp",
"need",
"network",
"performance",
"playwright",
"profile",
"reproduce",
"requests",
"runs",
"screenshots",
"testing",
"tests",
"uis",
"validate",
"validating",
"verify",
"web",
"webapp"
]
},
"platforms": [
"linux",
"darwin",
"windows"
],
"token-cost": 554,
"description": "ALWAYS invoke when testing a web application end-to-end in a real browser — driving flows, asserting UI, capturing console\/network\/performance evidence, and verifying visual output with real runtime data.",
"disable-model-invocation": false
}
Webapp Testing
Use When
- Reproducing frontend issues in a real browser.
- Validating local UI flows with Playwright.
- Capturing screenshots, console output, network observations, or rendered DOM state.
Do Not Use
- Looking up Playwright or browser API syntax without docs verification.
- Replacing
debug-investigationwhen the overall root cause is still unknown. - Owning frontend architecture, styling direction, or backend contracts.
Owns
- Browser-based UI reproduction and validation workflow.
- Safe local server lifecycle management for automation runs.
- Reconnaissance patterns for screenshots, console logs, network events, and DOM inspection.
Does Not Own
- The source of truth for Playwright or library APIs.
- Product debugging strategy outside browser evidence gathering.
- Implementation policy for app code changes.
Combine With
debug-investigationwhen browser evidence is one part of a broader bug investigation.code-execution-standardwhen the task includes fixing the issue after reproducing it.tool-and-doc-selectionwhen exact Playwright, browser, or test-runner syntax matters.
Workflow
- Confirm the workspace or environment already has the browser automation runtime you plan to use, then run
python3 scripts/with_server.py --helpbefore using the helper. - Start only the local services you need, and prefer explicit readiness checks such as
--ready-url, locator waits, or app-specific assertions over blanketnetworkidle. - Keep Playwright scripts focused on browser actions and assertions while the helper owns startup and teardown.
- Save screenshots and logs to workspace-safe temp locations such as
/tmpor a task-specific scratch folder inside the current workspace. - Route Playwright or browser-automation API questions through
tool-and-doc-selectionand the docs skills instead of treating this skill as API documentation.
Output Contract
- The local URLs or servers that were exercised and how readiness was verified.
- The browser evidence captured: screenshot paths, console output, network observations, or assertions.
- Any follow-up handoff to
debug-investigation,code-execution-standard, or docs skills.
版本历史
- 581d130 当前 2026-07-19 09:15
