e2e-automation-project
Maintain a dedicated automation test project per repository - separate from the product's unit tests - with API, worker, and UI suites written scenario-first
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-15
- Stars
- ★ 122
- Forks
- 18
01 前言 Front Matter
- name
- e2e-automation-project
- enabled
- false
- category
- testing
- description
- Maintain a dedicated automation test project per repository - separate from the product's unit tests - with API, worker, and UI suites written scenario-first
03 SKILL.md 正文
E2E Automation Project
Overview
Every repository worth testing gets a dedicated automation test project: a separate project that exercises the real product through its public surface and can be re-run on every change. It is not the product's unit-test folder and never mixes into it — it has its own dependencies, its own entrypoint, and runs in the repository's own pipeline (automation-pipeline-integration).
Core principle: scenarios before code (scenario-plan-first). A suite written without a scenario list tests whatever was easy, not what matters. And automation follows the manual pass: you automate the scenarios you have already executed and understood by hand.
Structure
One automation project per repository, following the project's convention — default qa-automation/ at the repo root (a sibling repo also works):
qa-automation/
scenarios/ # the human-readable scenario lists, one file per feature
api/ # backend API suite: HTTP client + assertions
worker/ # async job suite: trigger → poll → assert side effects
ui/ # frontend suite: Playwright against the built app
fixtures/ # seed data
support/ # helpers, page objects, request builders, stub mappings
scripts/e2e.sh # single entrypoint: provision, run all suites, teardown
| Surface | Tool (follow the project's existing choice) |
|---|---|
| REST API | the language's HTTP test client + assertions (Playwright APIRequest, RestAssured, supertest, Go net/http + testify) |
| Worker / async jobs | same test runner; triggers through the product path, polls for side effects (worker-job-testing) |
| Web UI | Playwright (preferred) / Cypress — headless Chromium via executablePath: process.env.CHROME_BIN, no browser download |
Mobile automation is out of scope for now — do not scaffold a mobile suite.
Determinism
- The app under test boots from the current branch build; the suite never targets stage, prod, or any shared environment.
- Database and infra are disposable containers seeded per run (test-database-seeding, Testcontainers or the project's equivalent).
- External services are stubbed and asserted on (test-doubles-wiremock).
- Each test seeds and cleans its own data; no ordering dependencies; re-running gives the same result.
The Workflow
- Author scenarios first for the feature: happy path, boundary, negative, auth/permission, key regressions — committed under
scenarios/and posted on the task. - Stand up the environment through the entrypoint script, never by hand-managed local state.
- Implement one test per scenario, named for the behavior (
export_forbidden_for_other_tenant), reading like the scenario via page objects / request helpers. - Run the whole suite, read the output, capture failure artifacts (screenshots, traces, logs) as evidence.
- Wire it into the pipeline if not already (automation-pipeline-integration) — a suite outside the pipeline enforces nothing.
Worked Example
Feature: task export to CSV.
- Scenarios: (1) export returns the project's tasks; (2) empty project → header-only CSV; (3) other tenant → 403; (4) export completion event is produced by the worker.
- Environment: app against a Testcontainers Postgres seeded with a known project; auth provider stubbed with WireMock.
- Tests:
api/export_returns_all_tasks,api/export_empty_project_header_only,api/export_forbidden_for_other_tenant,worker/export_event_emitted. Each seeds its own rows, asserts, cleans up. scripts/e2e.sh→ all green; pipeline runs the same script on the next task.
Common Mistakes
- Mixing automation into the product's unit-test folders — separate project, separate dependencies.
- A second automation project forked for the same product — extend the existing one.
- Suites that hit staging/shared data → flaky, order-dependent, forbidden.
- A scenario in the plan with no implementing test, or tests with no scenario.
Red Flags
- The suite passes locally but fails in a clean environment → hidden shared state.
- Re-running gives different results → nondeterminism to fix before anything else.
- UI tests download a browser at run time → use the system Chromium via
CHROME_BIN.
04 同仓库更多 Skills
visual-craft
暂无简介
ux-copy
Use when you write any words a user reads in a mockup or a hand-off - content language and voice, buttons, labels, errors, empty states, confirmations, numbers and dates, and the longest-string check
screen-mockup-html
Use when you draw screens as the design task's HTML mockup - ONE canvas document holding every screen and every variant side by side, the sanitizer-safe recipe, phone and wide frames, the states grid, static control states, and the loopback self-check with screenshots at 375, 768 and 1440 attached to the task
performance-awareness
How the performance score moves for your role. Use when the score context shows a declining trend, or before handing off work you have not self-checked against every criterion.
design-variants
Use when a screen design needs variants - how many, variants as side-by-side rows on the one canvas, what makes two variants genuinely different, recommending one, how the human's choice is read, and narrowing to the chosen one in revision
design-system-authoring
Use when you derive, change or extend a design system - where tokens live per stack, the DTCG token tree, role-based names for both themes, base vs layer vs drift, the DESIGN.md section order, the inventory format, the contrast check and the lint every proposal returns