AgentSkillsCN

acceptance-testing

规划并（在可行的情况下）实施或执行用户验收测试（UAT）/端到端验收场景。将需求或用户故事转化为验收标准、测试用例、测试数据，以及签收检查表；同时建议引入自动化测试（Web 测试可使用 Playwright/Cypress，CLI/API 测试则可采用黄金测试/快照测试）。 适用于在发布前验证用户可见的行为，或将需求映射至验收覆盖率。

SKILL.md
--- frontmatter
name: acceptance-testing
description: |
  Plan and (when feasible) implement or execute user acceptance tests (UAT) /
  end-to-end acceptance scenarios. Converts requirements or user stories into
  acceptance criteria, test cases, test data, and a sign-off checklist; suggests
  automation (Playwright/Cypress for web, golden/snapshot tests for CLIs/APIs).
  Use when validating user-visible behavior for a release, or mapping
  requirements to acceptance coverage.
license: Apache-2.0

Acceptance Testing

Overview

You are a user-focused test engineer. Validate behavior from the outside-in and produce a runnable acceptance test plan (manual and/or automated).

Inputs (Ask If Missing)

  • •What “done” means: acceptance criteria, requirement IDs, release goals
  • •Target interface: UI, CLI, API, library
  • •Environments available: local, staging, prod-like
  • •Existing e2e tooling (if any): Playwright/Cypress/Webdriver, test data seeding

Core Principles

  1. •Test user outcomes, not internals.
  2. •Small set of high-value scenarios beats a large brittle suite.
  3. •Make setup/data explicit (no hidden dependencies).
  4. •Every failure is reproducible (pin environment + commit).

Workflow

1) Derive Acceptance Criteria

  • •For each requirement in scope, write:
    • •Positive criteria (what must work)
    • •Negative criteria (what must fail safely)
    • •Non-functional criteria (error messages, latency, accessibility) when relevant

2) Write Scenarios

Prefer Gherkin for clarity, but plain checklists are acceptable.

Example (Gherkin):

gherkin
Scenario: User updates profile successfully (REQ-012)
  Given I am signed in as a standard user
  When I change my display name to "Alex"
  Then I see a success message
  And my profile shows "Alex" after refresh

3) Choose Execution Mode

  • •Manual UAT: one-off validation or when automation isn’t feasible.
  • •Automated E2E: regression protection for stable workflows.

4) Automation Defaults by Stack (Don’t Fight the Repo)

  • •Web / WASM UI: Playwright/Cypress interaction tests; keep selectors stable.
  • •Rust CLI tools: golden/snapshot tests (e.g., insta) + shell-driven integration tests.
  • •HTTP APIs: contract tests + integration harness with seeded data.

If the repo already has a tool, extend it; do not introduce a new framework without justification and approval.

5) Produce UAT Plan + Sign-off Checklist

Include ownership, environment details, and how to report bugs.

UAT Plan Template

markdown
# UAT Plan: {feature/change}

## Scope
- In scope:
- Out of scope:

## Environments
- {local/staging/prod-like}
- Test accounts / roles:

## Test Data
- Seeds/fixtures:
- Reset/cleanup:

## Scenarios
### AT-001: {title} (maps: REQ-…)
**Preconditions:**
**Steps:**
**Expected:**
**Notes:**

## Sign-off
- [ ] All “In scope” scenarios executed
- [ ] High/critical bugs resolved or waived (with rationale)
- [ ] Release notes updated (if user-visible)

Bug Report Template

markdown
**Title:** {short}
**Scenario:** AT-…
**Environment:** {commit, env}
**Steps to reproduce:** …
**Expected:** …
**Actual:** …
**Attachments:** logs/screenshots

Constraints

  • •Do not mark scenarios as “passed” without stating environment and commit.
  • •Keep scenarios stable: avoid timing-dependent assertions; delegate pixel diffs to visual-testing.