Token导航 LogoToken导航TokenDH.com
研究检索权限需确认github未标认证来源可访问许可证需确认审计通过

test测试

Agent Skill

用于辅助测试设计、自动化测试、用例整理和回归验证。它适合让 Agent 编写单元测试、端到端测试、测试计划或根据失败日志定位问题。使用时需要确认项目测试框架、运行命令和夹具数据,避免为了通过测试而改坏真实逻辑;涉及浏览器或外部服务时,应区分本地模拟、测试环境和生产环境。

总安装

343

周安装

14

GitHub Stars

18

下载量

111
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:test(测试)
来源仓库:https://github.com/jay-523/agent-skills
仓库路径:skills/test
安装命令:
npx skills add https://github.com/jay-523/agent-skills --skill test
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/jay-523/agent-skills --skill test

简介

test 用于辅助测试设计、自动化测试和用例整理。

  • 适合让 Agent 编写单元测试、端到端测试或根据失败日志定位问题。
  • 使用时需确认项目测试框架、运行命令和夹具数据。test 属于研究检索类 Skill,可作为该场景下的辅助能力补充。
  • 涉及浏览器或外部服务时,应区分本地模拟与生产环境。
  • 避免为了通过测试而改坏真实逻辑,确保改动符合业务预期。

SKILL.md

/test -- Full TDD Red-Green-Refactor Cycle

Run a complete test-driven development cycle for the given task. Four phases: Discovery, Red, Green, Refactor, then final Verification.

Argument

A description of the feature or behavior to implement via TDD.

Procedure

Phase 0 -- Discovery

  1. Detect the test framework by checking config files:

- pyproject.toml, setup.cfg, pytest.ini for Python (pytest, unittest) - package.json for JS/TS (jest, vitest, mocha) - Cargo.toml for Rust - go.mod for Go - Fall back to asking the user if ambiguous

  1. Identify the test command (e.g., pytest, npm test, cargo test)
  2. Read 2-3 existing test files to learn project conventions:

- File naming pattern (test_*.py, *.test.ts, etc.) - Import style, fixture usage, assertion style - Directory structure (tests/, __tests__/, co-located)

  1. Confirm scope with the user before proceeding:

- "I will write tests for [X] using [framework] in [directory]." - "Test cases planned: [list]" - Wait for user approval.

Phase 1 -- RED (Write Failing Tests)

Write comprehensive tests covering:

  • Happy path (expected inputs produce expected outputs)
  • Edge cases (empty inputs, boundary values, type variations)
  • Error cases (invalid inputs, expected exceptions)
  • Integration points (if the feature interacts with other modules)

Conventions to follow:

  • Add docstrings to every test function explaining what it validates
  • Use descriptive test names (test_returns_empty_list_when_no_matches)
  • No emojis anywhere
  • Follow existing project test patterns discovered in Phase 0

Run the tests:

source venv/bin/activate && [test command] [test file]

All tests MUST fail. If any test passes, it is not testing new behavior -- fix or remove it. A passing test in the RED phase means the test is not validating anything new.

Phase 2 -- GREEN (Minimum Implementation)

Implement the minimum code to make each test pass:

  • Add docstrings with the algorithm in plain English, args, and input/output documentation
  • For functions taking loose objects (dict, pandas DataFrame, list, etc.), write the expected schema explicitly in the docstring
  • Add if __name__ == '__main__' blocks with hardcoded example inputs (no argparse)
  • Use uv pip install for any new packages (never bare pip)
  • Always source venv/bin/activate before running anything

After each implementation change, run the tests:

source venv/bin/activate && [test command] [test file]

If tests still fail, iterate. Maximum 10 attempts before stopping and reporting what is stuck. Do not loop infinitely.

Phase 3 -- REFACTOR

With all tests passing, clean up:

  • Remove duplication in implementation code
  • Improve variable/function naming for clarity
  • Extract helpers only where they reduce genuine complexity (not for one-time operations)
  • Ensure docstrings are accurate after changes

After EACH refactor step, run the tests:

source venv/bin/activate && [test command] [test file]

If any test fails after a refactor, revert that specific change immediately. Refactoring must not change behavior.

Phase 4 -- Verification

  1. Run the full test suite (not just new tests): source venv/bin/activate && [test command]
  2. Run the project linter if one is configured:

- Check pyproject.toml for ruff/flake8/black config - Check package.json for eslint/prettier scripts - Run it and fix any issues in new code only

  1. Report final status: ## TDD Cycle Complete Tests written: [count] Tests passing: [count] Implementation files: [list] Linter status: [pass/fail/not configured] ### Test Coverage - Happy path: [covered/not covered] - Edge cases: [covered/not covered] - Error cases: [covered/not covered] - Integration: [covered/not covered]

Important

  • Never skip the RED phase. Tests must fail first to prove they test real behavior.
  • Never skip running tests between changes. The cycle depends on continuous feedback.
  • If the task is ambiguous, ask for clarification in Phase 0 before writing any tests.
  • Keep test files focused. One test file per module/feature being tested.
  • Do not refactor code outside the scope of this task.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.47%
按下载量换算38

Claude

31.11%
按下载量换算35

Cursor

16.43%
按下载量换算18

Gemini CLI

9.13%
按下载量换算10

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

权限需确认

当前来源未能明确判断权限范围,默认进入异常复核队列。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills