AGENT SKILL

browser-qa

Playwright-based browser QA skill for guided setup, auth handoff, scripted runs, and artifact capture.

使用场景测试与质量
License 许可
AGPL-3.0
最近更新
2026-08-15
Stars
★ 30
Forks
7

01 前言 Front Matter

name
browser-qa
description
Playwright-based browser QA skill for guided setup, auth handoff, scripted runs, and artifact capture.
allowed-tools
Read, Grep, Glob, Bash, ApplyPatch, Question
argument-hint
[setup|start|auth|inspect|run|qa-script|report] [--app <path>] [--url <url>] [--script <path>] [--headed] [--allow-destructive]
disable-model-invocation
true

02 资源文件 6

03 SKILL.md 正文

Browser QA

Purpose

Use Playwright for guided browser-based QA across local, preview, staging, or production environments while keeping setup deterministic, credentials safe, and artifacts easy to review.

Inputs

  • Optional subcommand: setup, start, auth, inspect, run, qa-script, or report

  • Optional flags: --app <path>, --url <url>, --script <path>, --headed, --allow-destructive

  • Template paths are resolved from ${AGENTS_SKILLS_ROOT}/browser-qa/.

Steps

  1. Follow the command specification in ${AGENTS_SKILLS_ROOT}/browser-qa/commands/browser-qa.md exactly.
  2. Use ${AGENTS_SKILLS_ROOT}/browser-qa/scripts/detect-runtime.js to inspect README.md, app-level READMEs, and package metadata before asking startup questions.
  3. Use ${AGENTS_SKILLS_ROOT}/browser-qa/scripts/check-playwright.js to deterministically report whether Playwright packages and browser binaries are ready before attempting browser execution.
  4. Default to read-only mode. Only perform create/edit/delete/submit/approve side effects when the user explicitly opts into destructive actions.
  5. Default to manual login in a headed browser for first-run auth unless the user explicitly chooses typed credentials, env-backed credentials, or an existing storage state.
  6. Never persist secrets or session state unless the user explicitly asks. Redact sensitive values from summaries.
  7. Use ${AGENTS_SKILLS_ROOT}/browser-qa/scripts/parse-qa-script.js to normalize markdown or JSON QA scripts into a deterministic scenario list before execution.
  8. Use ${AGENTS_SKILLS_ROOT}/browser-qa/scripts/summarize-artifacts.js after execution to normalize artifact paths, scenario outcomes, and blocked reasons for handoff to the user or the qa skill.
  9. When creating or updating a reusable Markdown QA script, also export a sibling .xlsx workbook with python3 ${AGENTS_SKILLS_ROOT}/qa/scripts/export-user-test-script.py <markdown-path> [xlsx-path] so human testers can use either format.
  10. When browser execution is requested from the broader qa workflow, return a structured summary that qa can incorporate into bug reports and QA artifacts.

Output

  • Friendly browser-QA summary with target app/URL, auth mode, safety mode, actions performed, artifact locations, and pass/fail/blocked results.
  • Clear list of missing prerequisites when Playwright, browsers, or startup commands are not ready.
  • Structured scenario summary suitable for qa delegation and follow-up fixes.

04 同仓库更多 Skills