用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/aibot88/sec_skill_store --skill authoring-tests命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | authoring-tests |
| description | Use when writing, reviewing, or modifying any test files, or when asked to add test coverage. |
| paths | ["**/*.test.{js,cjs,mjs,jsx,cjsx,mjsx,ts,cts,mts,tsx,ctsx,mtsx}","**/*.spec.{js,cjs,mjs,jsx,cjsx,mjsx,ts,cts,mts,tsx,ctsx,mtsx}","tests/**/*","testing/**/*"] |
| user-invocable | false |
Adhere to these principles when writing tests:
Use arrange-act-assert:
Keep each test focused and orthogonal to other tests unless you're testing an interaction between two behaviors or each test is expensive.
Prefer parameterized testing over multiple independent calls and assertions in the same test.
Test behaviors, not functions. A single function may exhibit many behaviors, and a single behavior sometimes spans across multiple functions.
Write test names that describe the scenario and expected outcome of the test.
Avoid overspecifying tests. Rule of thumb: if changing the expected data in an assertion does not change the meaning of test, then it shouldn't be asserted.
Don't test implementation details. Test the public API only.
Avoid mocks. Prefer real implementations and test doubles over mocks. Reserve mocks for I/O boundaries and third-party services. If you must use mocks, then ONLY mock public APIs.
Extract well named variables to split up and clarify complex test data.
Use helper functions to remove redundant details from the test body. NEVER hide details relevant to the test in helper functions. Pass the relevant data into the helper function from the test instead.
Don't put logic in tests. Avoid conditionals and loops, and state inputs and outputs directly to avoid bugs in the tests themselves. If the logic is truly necessary, then extract it to a function and test it as well.
Do not write redundant change detector tests. If the test would break for any change in the code under test, then it's likely a change detector test.