# Unit tests that give false comfort

Prompt · Testing and security · T01

Finds features whose automated checks all pass and that still break for real people.

## Prompt

```text
Find critical user journeys that have extensive unit-test coverage but insufficient evidence that they work end to end. Strengthen the smallest set of tests that closes the real gaps.
```

**What it returns:** Each journey with what the tests cover, what they skip (real auth, real DB, real browser), and the one added check.

**Skip it when:** User paths already covered by a browser test.

**Tools:** Playwright or the in-app browser for the full start-to-finish check.

**Changes:** Adds a test that shows it

**Sessions:** One session

## Skills

- [tdd](https://atlasscope.dev/skills/tdd): Test-driven development.
- [viewport-proof](https://atlasscope.dev/skills/viewport-proof): Use only when Austin explicitly invokes /viewport-proof or $viewport-proof.

## Related

- [Tests that can't fail](https://atlasscope.dev/prompts/tests-that-cant-fail): Finds tests that pass no matter what.
- [What the tests really prove](https://atlasscope.dev/prompts/what-the-tests-really-prove): Asks what each test really proves, then adds one check that would catch a real break.
- [Why production might not match main](https://atlasscope.dev/prompts/why-production-might-not-match-main): Lists every way 'merged' can fail to mean 'live'.

Source: https://atlasscope.dev/prompts/unit-tests-that-give-false-comfort
