AgentSkillsCN

gen

将OpenSpec功能规格转换为确定性的Playwright测试，采用稳定的选择器与显式的断言。

SKILL.md
--- frontmatter
name: gen
description: Convert an OpenSpec feature spec into deterministic Playwright tests using stable selectors and explicit assertions.
disable-model-invocation: true
argument-hint: "[path-to-openspec.md] [optional: output-dir]"
allowed-tools: Read, Write, Grep, Glob, Bash(node *), Bash(npx playwright *)

Goal

Given an OpenSpec file path, generate Playwright tests that model user flows as test cases, use stable selectors, assert real acceptance criteria, and avoid brittle timing.

Inputs

  • •OpenSpec file: $ARGUMENTS[0]
  • •Output directory (optional): $ARGUMENTS[1], default tests/ui/generated

Output

  • •One spec file per OpenSpec "User Flow" (or equivalent).
  • •Write tests to: <output-dir>/<slug>.spec.ts
  • •If needed, write shared helpers under <output-dir>/_helpers/ (page objects, login helper).

Core Workflow

1. Parse the OpenSpec

Follow the extraction procedure in references/parsing-rules.md:

  1. •Feature name / title (top heading)
  2. •Preconditions / Setup (logins, seed data)
  3. •User Flows (numbered steps)
  4. •Acceptance Criteria (assertions)

If any section is missing, infer minimally and add a TODO comment in the test file.

2. Generate test files

Apply the selector strategy and structure patterns in references/test-writing-rules.md. Use the scaffold in templates/playwright.spec.ts.template as baseline structure.

See references/example-openspec.md for a worked example.

3. Post-generation checks

  1. •If playwright.config.* exists, align baseURL usage.
  2. •Run a quick syntax check:
    bash
    npx playwright test <generated-file> --reporter=line
  3. •If it fails due to missing app/server, do NOT start the server automatically unless the OpenSpec explicitly states the command.

Final Response

  • •List created/modified files.
  • •Provide the command to run the generated suite.
  • •Call out any TODOs introduced.