Skip to main content

write-tests

Write colocated tests for React + Express + TypeScript code using the project's runner, conventions, mocked repository interfaces, React Testing Library, and TanStack Query patterns.

来源信息

仓库
ThibautBaissac/react-express-ai-kit
最近来源活动
2026年6月9日 07:28
检测到的 SKILL.md 语言
英语
星标
2
Forks
0

安装方式

默认使用会先检查来源的 Prompt;你也可以切换为直接命令,或下载本地副本。

检查来源文件

决定是否安装前,请先阅读 SKILL.md,以及 SkillsMP 当前展示的配套文件。

文件资源管理器
2 个文件

正在显示 SKILL.md

SKILL.md
来源说明 · 只读预览
name
write-tests
description
Write colocated tests for React + Express + TypeScript code using the project's runner, conventions, mocked repository interfaces, React Testing Library, and TanStack Query patterns.
when_to_use
Use WHEN adding tests, improving coverage, or testing a module, service, hook, component, route, or feature — whether the user asks directly or an implementation agent is realizing a Testing To-Do item from a `tasks/task-N.md` plan. Do NOT use to run an existing suite (use /run-checks), and do NOT use for browser end-to-end flows (those are Playwright specs in `e2e/` — use /e2e).
argument-hint
[path-to-test]
arguments
["target"]
model
sonnet
# Write tests to convention Test `$target`. Colocate tests. ## Step 1 — Detect runner and style ```bash bash "${CLAUDE_PROJECT_DIR}/.claude/lib/detect-toolchain.sh" ``` Open nearby tests first. Match imports, setup, mocks, naming, and `references/test-patterns.md`. ## Step 2 — Pick the test type - Service tests use a mocked repository interface and no DB or HTTP. - Repository tests are integration tests only when the project already has a test DB harness. - Query-backed hooks and components use React Testing Library with a fresh QueryClient and `retry: false`. - Presentational components render with props and assert by role, label, or text. - Route tests use the project's existing supertest-style pattern when one exists. ## Step 3 — Write behavior tests Use Arrange-Act-Assert. Test observable outcomes, including meaningful error or edge cases, not private internals. ## Step 4 — Run new tests ```bash source "${CLAUDE_PROJECT_DIR}/.claude/lib/detect-toolchain.sh" run_tests <new-test-path> ``` Iterate until tests pass. Report covered behaviors, not a coverage percentage. ## Checklist - [ ] Runner and local test style were detected. - [ ] Test file is colocated with the code under test. - [ ] Services use a mocked repository interface. - [ ] Query-backed UI uses a fresh QueryClient with `retry: false`. - [ ] Assertions target user-visible behavior or boundary calls. - [ ] New tests pass through the detected runner.
在 GitHub 查看