e2e
End-to-end checks in a headless browser — scripted Playwright tests for main paths, and an exploratory run that turns findings into tests. Use for "прогони e2e", "проверь в браузере", "e2e", "test the UI".
- License 许可
- MIT
- 最近更新
- 2026-11-08
- Stars
- ★ 3
- Forks
- 1
01 前言 Front Matter
- name
- e2e
- description
- End-to-end checks in a headless browser — scripted Playwright tests for main paths, and an exploratory run that turns findings into tests. Use for "прогони e2e", "проверь в браузере", "e2e", "test the UI".
03 SKILL.md 正文
End-to-end in a real browser
E2E tests assert what the user gets, never that a selector exists (D049).
Triggers
- RU: «прогони e2e», «проверь в браузере», «пройди сценарий»
- EN: "run e2e", "check it in the browser", "walk the flow"
0. Tools
Scripted tests need the project's Playwright dev dependency. Exploratory mode needs the
Playwright MCP server. Both are installed only on the owner's word. Without them, say what is
missing and stop here.
Done when: npx --no playwright --version runs (--no keeps npx from downloading it), or
the gap is reported.
1. Scripted tests (in the project, run in CI)
- Cover login and the main user paths first.
- Locators by role or test id, never by CSS structure.
- Assert outcomes: the saved record, the total shown, the file downloaded.
- Clean state per run: a test user, fresh data, stored auth from a setup project.
- A trace on failure (
trace: "retain-on-failure"). - Run through
bedrock cmd e2e, so the heavy-run lock applies. Done when: the new test fails before the change it guards and passes after.
2. Exploratory run (Playwright MCP)
- Start from a clean state with a test user.
- Walk the flows the plan names. Try odd input, the back button, a double submit, a slow network, a narrow window.
- For UI work: screenshots at 375, 768 and 1280 px; the flow by keyboard only; an axe scan.
- Every break found becomes a scripted test from step 1, then goes to the
debugskill. Done when: each finding has a scripted test or a reason it cannot have one.
3. Report
What was checked, on which environment, with what result: the command and its output, or the screenshot paths. A check that did not run is named as not run. Done when: every flow in the plan has a result or a "not run" line.
04 同仓库更多 Skills
writing
Write or rewrite skills, agent definitions, briefs and rule files by the D013 rules, then check them. Use for "напиши скилл", "поправь агента", "write a skill", "edit this brief".
triage
Bug triage — reproduce each report on the current version, group duplicates, set priority, decide fix, defer or decline. Use for "разбери баги", "что с репортами QA", "triage", "new bug reports".
start
Start a project with bedrock — a new one by interview, or an existing one by onboarding; writes the base, AGENTS.md, workspace settings and verified commands. Use for "новый проект", "подключи проект", "start a project".
review
Review gate for a finished block — deterministic checks, code review, plan-compliance review, silent fixes, rounds until clean. Use for "проверь", "ревью", "review this", before any PR.
retro
Weekly retro — read the week's logs, agents, reviews and failures, find what repeats, tune the system, plan a realistic next week. Use for "ретро", "итоги недели", "retro", "how did the week go".
night
Autonomous work until morning — a pre-sleep check, the allowed list, auto-continue with fuses, a watchdog, a morning report. Use for "автономно до утра", "поработай ночью", "work overnight".