AgentSkillsCN

testing

在更改前后运行测试。只添加复杂逻辑的测试，不要添加样板。数据库测试最重要。

SKILL.md
--- frontmatter
name: testing
description: >-
  Run tests before and after changes. Only add tests for complex logic,
  not boilerplate. Database tests matter most.

Testing

Verify changes don't break things. Don't create test bloat.

Announce: "I'm running tests to verify the changes."

Philosophy

Database (pgTAP): This is where real logic lives. Test it thoroughly.

Frontend (Vitest): Thin RPC layer. Only test:

  • •Complex transformations
  • •Non-obvious edge cases
  • •NOT: "does this call that function"

Workflow

Before implementing:

bash
bun run test

If tests fail, fix them first.

After implementing:

bash
bun run test

If tests fail, your change broke something. Fix it.

When to Add Tests

YES - add tests for:

  • •New database functions with business logic
  • •Complex data transformations
  • •Non-obvious edge cases you discovered

NO - don't test:

  • •"Does this function call that function"
  • •Pinia store initialization
  • •Vue component mounting
  • •API calls with mocked responses
  • •Obvious behavior readable from code

Commands

bash
bun run test        # All tests (unit + db)
bun run test:unit   # Frontend only
bun run test:db     # Database only (pgTAP)

If Tests Fail

  1. •Read the error message
  2. •Check if YOUR change caused it
  3. •If yes → fix your code
  4. •If no → fix the test or report the issue