> ## Documentation Index
> Fetch the complete documentation index at: https://yoqa.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Test Cases

> Use natural language instructions for the AI agent.

## Where to create

In the desktop app: open an app → **Test Cases** → add a case. Or use the CLI:

```bash theme={null}
yoqa cases list APP
yoqa cases create APP '{"title": "Login", "tags": ["smoke"], "flows": [{"instructions": "Sign in with the test account", "result": "Home screen is visible"}]}'
```

## Structure

A test case has a **title**, one or more **flows**, and optional **tags**.

### Flows

* **Instructions** — natural language description of what the agent should do.
* **Expected result** — what to verify after the instructions. If omitted, the agent mainly checks that the flow completed without errors.
* **Reusable flows** — reference a shared flow by id instead of inlining instructions. Update the reusable flow once; every case that uses it picks up the change.

Each flow is either inline (`instructions` + `result`) **or** a reusable reference (`id`) — never both.

### Tags

Tags (e.g. `smoke`, `regression`) help filter cases in the UI and CLI (`yoqa cases list APP --tag smoke`). Naming a new tag creates it automatically.

## Configuration

Per-case options:

* **Appium capabilities** — override how the session starts for this case only. See [Appium capabilities](/guide/best-practices-appium-capabilities).
* **Saved scripts** — after a passing AI run, you can save a scripted replay for faster subsequent runs (script vs AI mode).

<Tip>
  For phrasing tips, see [Writing good test cases](/guide/writing-test-cases).
</Tip>