> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.qa.army/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.qa.army/_mcp/server.

# How QA.army Works

> See how QA.army turns important product journeys into readable tests and clear results.

QA.army brings AI into the full QA workflow while keeping coverage readable and under your team's control.

> **Warning**
>
> Direct Web testing is available through the SaaS app and authorized CLI,
> local MCP, or Product API operations. Native Android, iOS, and Desktop execution
> are not generally available. Each interface uses the same saved Tests and Run
> results.

## 1. Define what matters

Start with the journeys whose failure would hurt customers or block a release. Describe each journey as plain-language actions and observable outcomes.

```text
Open the sign-in page.
Sign in with the test account.
Verify the account dashboard is visible.
```

Save related journeys in a Test Group so your team can review coverage by product area.

## 2. Choose where to test

Add environments such as Production, Staging, and QA to the Project. The same saved Test can target different approved environments without duplicating its steps.

## 3. Run with clear permissions

Choose the Test and Environment, then request a Run through an available
customer interface. Workspace permissions control who can manage Tests, Runs,
and evidence.

## 4. Understand the outcome

Every completed Run should answer one question: did the product journey work?

* `PASSED` means every required assertion succeeded.
* `FAILED` means the journey ran and a product assertion failed.
* `ERROR` means QA.army could not reach a reliable product conclusion.
* `CANCELLED` means the Run stopped before completion.

When evidence is available, use screenshots, recordings, activity, and reproduction steps to understand what happened.

## Keep control

Review generated coverage before saving it. Use dedicated test accounts, avoid real customer data, and enable only the permissions required by the journey.

## Availability

Use the SaaS app, published CLI, local MCP, or Product API for authorized
saved Web Test and Run operations. Review the actual result and available
screenshots and recordings; missing evidence is not a passing result. Native
Android, iOS, and Desktop execution and scheduled monitoring are not generally
available. Dynamic pull-request testing is an assisted pilot and is not
generally available.