Unit Test Writer
Writes focused unit tests for a function or module — covering the happy path, edge cases, and error handling — in the project's existing test framework.
What this skill does
Key features
- Detects and matches the project's existing test framework and style
- Covers happy path, edge cases, and error handling deliberately
- Names tests by behaviour, not implementation
- Doesn't test private internals or over-mock
Use cases
- Backfilling tests for untested code
- Adding edge-case coverage to a fragile function
- Scaffolding tests in TDD before implementing
SKILL.md
The skill definition — its metadata and the exact instructions an agent follows. Copy it into a SKILL.md file in your agent's skills folder.
| name | test-writer |
|---|---|
| description | Write focused unit tests for a target function using the project's test framework. |
| allowed-tools | Read, Grep, Bash(ls:*) |
| coverage | happy path + edges + errors |
Musts
- Detect the existing test framework (look for jest/vitest/pytest config and existing test files) and match its style.
- Cover, deliberately: the happy path, boundary/edge cases, and error/invalid-input handling.
- Name each test by the behaviour it verifies (
returns 0 for an empty list), not the implementation. - Test observable behaviour through the public API — don’t reach into private internals.
Guidelines
- Mock external I/O (network, filesystem, clock) but avoid over-mocking pure logic.
- One assertion focus per test; use clear arrange/act/assert structure.
- If a function is hard to test, say why (it likely needs refactoring) rather than writing a brittle test.
Example (Jest)
describe('slugify', () => {
it('lowercases and hyphenates spaces', () => {
expect(slugify('Hello World')).toBe('hello-world');
});
it('returns an empty string for empty input', () => {
expect(slugify('')).toBe('');
});
});FAQ
Does it run the tests?
It writes them; running is a separate step you control. It will note how to run them (e.g. `npm test`) but won't execute anything on its own here.
Will it reach 100% coverage?
Coverage is a means, not a goal. It targets meaningful behaviours and edge cases; chasing 100% often just tests trivia. Review what it wrote.
Before running any skill, read its instructions and the tools it's allowed to use, and test it on a safe target first. See LLM security best practices.