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

# Review and approve AI test cases

> Review, edit, and approve AI-drafted test cases in TestFactors — the specific situations within each scenario that your team needs to verify.

<Note>
  **Test cases are where 'what we want to test' becomes 'exactly what we'll verify.'** This is the layer that catches the edge cases that scenarios are too broad to cover.
</Note>

## What this module does

A **scenario** is broad ("Hire a US employee into an active position"). A **test case** is a specific instance of that scenario with concrete data:

> *Scenario:* Hire a US employee into an active position
>
> *Test case 1:* Hire a salaried US employee, exempt, with stock eligibility, into an existing active position in the Engineering department.
>
> *Test case 2:* Hire an hourly US employee, non-exempt, no stock, into a position requiring background check.
>
> *Test case 3:* Hire a US employee into a position with future-dated start date (Q3 next year).

TestFactors generates test cases from your scenarios using:

1. **Workbook data** — every field in your spec that affects the scenario becomes a variable.
2. **Picklist values** — every option in a relevant picklist becomes a variant.
3. **Edge case patterns** — null values, max-length strings, boundary dates, special characters.
4. **Your knowledge wiki** — any custom rules or constraints you've documented.

You stay in control: every AI-generated test case sits in a **Proposed** state until you approve, edit, or reject it.

***

## Who uses this module

| Role                                        | Generate  | Approve / Promote | Edit    | Reject |
| ------------------------------------------- | --------- | ----------------- | ------- | ------ |
| **Test Manager / Project Manager**          | Yes       | Yes               | Yes     | Yes    |
| **Test Coordinator / Module Process Owner** | Yes       | Yes               | Yes     | Yes    |
| **Tester**                                  | Yes (own) | No                | Limited | Yes    |
| **Developer**                               | Yes (own) | No                | Yes     | Yes    |
| **Viewer**                                  | No        | No                | No      | No     |

<Tip>
  Testers can propose test cases and reject AI output, but only Test Coordinator+ can *promote* a proposed test case to active. This keeps the active test library curated.
</Tip>

***

## Before you start

You need:

* **At least one approved scenario** in this project. See [Scenarios & Categories](/platform/scenarios-and-categories).
* **A test cycle selected.** Test cases are generated per cycle, just like scenarios.

***

## Step-by-step: generate and review test cases

<Steps>
  <Step title="Open a scenario">
    From the Scenarios tab, click any active scenario. You land on the **Scenario detail view** with a list of its test cases (empty on first open).

    <Frame caption="Scenario detail page with an empty test cases section and a 'Generate Test Cases' call-to-action.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/01-scenario-detail.png" alt="Scenario detail empty" />
    </Frame>
  </Step>

  <Step title="Click Generate Test Cases">
    Click the primary **Generate Test Cases** button. The AI starts analyzing the scenario plus the underlying workbook data.

    A progress banner appears: *"AI is drafting test cases…"*

    <Frame caption="Generation in progress with progress banner and the first few proposed test cases starting to appear.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/02-generating.png" alt="Test case generation in progress" />
    </Frame>
  </Step>

  <Step title="Wait for proposals">
    Generation usually takes 15–30 seconds depending on scenario complexity. You'll see between 3 and 25 proposed test cases per scenario.

    Each proposal arrives with:

    * A **clear title** ("Hire salaried exempt US employee with stock")
    * **Preconditions** ("Position must be Active, requisition approved")
    * **Test steps** (numbered actions)
    * **Expected results** (one per step)
    * **Test data** (the specific values to use)

    <Frame caption="Test cases list showing 12 proposed cases in 'Proposed' state with a Review badge on each.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/03-proposed-list.png" alt="Proposed test cases list" />
    </Frame>
  </Step>

  <Step title="Open a proposal and review">
    Click any proposed test case to open the detail panel. You'll see the full structure: title, description, preconditions, steps with expected results, test data, and AI rationale ("Why did the AI propose this?").

    <Frame caption="Test case detail panel with steps, expected results, test data, and AI rationale visible.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/04-test-case-detail.png" alt="Test case detail" />
    </Frame>
  </Step>

  <Step title="Approve, edit, or reject">
    Each proposal has three primary actions in the top-right:

    * **Approve** — promotes the proposed test case to **Active** in your test library
    * **Edit** — open inline editor to tweak title, steps, expected results, or test data, then approve
    * **Reject** — discards the proposal. Optionally add a one-line reason that trains the AI for next time

    <Frame caption="Test case action bar showing Approve, Edit, Reject buttons and the 'Approval feedback' field.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/05-action-bar.png" alt="Test case action bar" />
    </Frame>

    <Tip>
      **Don't edit unless you must.** If the AI's title or steps are close-enough, approve as-is — you can always refine later. Most teams approve 70–80% of proposals without changes.
    </Tip>
  </Step>

  <Step title="Bulk approve when you're confident">
    For tight scenarios with consistent quality proposals, use **Bulk Approve**: select multiple proposals with the checkboxes and click **Approve selected**.

    <Frame caption="Test cases list with 5 checkboxes ticked and the 'Approve selected (5)' bulk action visible.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/06-bulk-approve.png" alt="Bulk approve" />
    </Frame>
  </Step>
</Steps>

<Check>
  **You're done.** Approved test cases are now in your test library. The next step is generating **test scripts** — step-by-step instructions a tester can follow. See [Test Scripts](/platform/test-scripts).
</Check>

***

## Refining proposals with chat

Like every AI surface in TestFactors, test case generation has a chat panel. After the first batch of proposals lands, try:

> *"Add 3 more negative test cases focused on validation failures"*
>
> *"The data values are too generic — use actual values from our workbook for the Job Code field"*
>
> *"Split the long test case in step 4 into two smaller cases"*
>
> *"Make the test data more realistic — use real-looking US names and SSN-style IDs"*

<Frame caption="Chat panel open beside the test cases list with a user request and the AI's proposed changes highlighted.">
  <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/07-refinement-chat.png" alt="Refinement chat" />
</Frame>

The AI's revised proposals appear as new "Proposed" rows. Your previously approved test cases are never touched.

***

## Editing a test case

<Steps>
  <Step title="Open the test case">
    Click any test case (proposed or active) to open the detail panel.
  </Step>

  <Step title="Click Edit">
    Top-right of the detail panel. The view becomes inline-editable.

    <Frame caption="Test case in edit mode with all fields editable inline.">
      <img src="https://mintlify.s3.us-west-1.amazonaws.com/testfactorsllc/images/test-cases/08-edit-mode.png" alt="Test case edit mode" />
    </Frame>
  </Step>

  <Step title="Make your changes">
    * **Title** — change wording
    * **Description** — add background
    * **Preconditions** — add or remove
    * **Steps** — add, reorder (drag-and-drop), delete
    * **Expected results** — one per step
    * **Test data** — change values, add columns, remove columns
  </Step>

  <Step title="Save">
    Click **Save** in the top-right. The status stays as it was (Proposed stays Proposed, Active stays Active). Edits are versioned — you can always see what changed in the **History** tab.
  </Step>
</Steps>

***

## Tips

<CardGroup cols={2}>
  <Card title="Approve in waves" icon="waves">
    Approve the obvious 70% first, then come back to edge cases. Reviewing the easy ones fast frees mental energy for the hard ones.
  </Card>

  <Card title="Use bulk approve aggressively" icon="check-double">
    If the first 5 proposals are good, the next 10 from the same batch usually are too. Trust the pattern.
  </Card>

  <Card title="Reject with one-line reasons" icon="comment">
    *"Wrong job code — we don't use Engineering"* is gold for AI training. *"Bad"* is not. Specifics make the next batch noticeably better.
  </Card>

  <Card title="Edit, don't reject + recreate" icon="pen">
    If 90% of a proposal is right and 10% is wrong, edit. Rejecting and asking the AI to re-propose costs you tokens and re-review time.
  </Card>

  <Card title="Group related cases" icon="layer-group">
    Test cases inherit from their scenario, which inherits from a category. Keep this hierarchy clean and your test runs will be much easier to navigate.
  </Card>

  <Card title="Don't manually write what AI can draft" icon="ban">
    Even if you "know" what the test cases should be, generate first. The AI catches edge cases you'd skip when typing them by hand.
  </Card>
</CardGroup>

***

## Test case states

| State        | Meaning                                            | Can be executed? |
| ------------ | -------------------------------------------------- | ---------------- |
| **Proposed** | AI drafted, awaiting human review                  | No               |
| **Active**   | Approved, in the live test library                 | Yes              |
| **Edited**   | Modified after approval                            | Yes              |
| **Archived** | Removed from active use but kept for history       | No               |
| **Rejected** | Proposal was rejected; reason kept for AI training | No               |

State badges appear next to every test case in lists.

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="The AI proposed identical-looking test cases" icon="copy">
    Usually means your scenario is too narrow — there's only one meaningful way to test it. Two options:

    1. **Broaden the scenario** — open the scenario, edit the title/description to cover more ground, regenerate
    2. **Approve the best one, reject the rest** with reason *"Duplicate of TC-XXX"* — trains the AI not to duplicate next time
  </Accordion>

  <Accordion title="The Approve button is missing" icon="lock">
    Your role doesn't include test case promotion. Tester role can propose and reject but not approve. Ask your Test Coordinator or Project Manager to approve, or to grant you a higher role.
  </Accordion>

  <Accordion title="Generated test data has placeholder values like 'TBD' or 'XXX'" icon="circle-question">
    The AI uses placeholders when it can't infer a concrete value from your workbook. Two fixes:

    1. **Edit the placeholder before approving** — replace with your actual test data
    2. **Add the data to your knowledge wiki** — document standard test values for your project so the AI uses them on the next generation
  </Accordion>

  <Accordion title="I want to mass-reject everything from one batch and re-generate" icon="trash">
    Select all checkboxes → **Reject selected**. Then click **Generate Test Cases** again. You can pass guidance in the chat: *"Don't focus on validation — generate happy-path cases only this time."*
  </Accordion>

  <Accordion title="Generation never finishes (banner stuck)" icon="hourglass-half">
    Typical generation finishes in under 60 seconds. If it's stuck:

    1. Refresh the page — proposals may already be there
    2. Click **Cancel generation** (if visible)
    3. Retry; if it stalls again, contact your Admin
  </Accordion>

  <Accordion title="I edited a test case but the changes aren't reflected in the test script" icon="link-slash">
    Test scripts are generated from test cases at a specific point in time. If you edit the test case after a script was generated, **regenerate the script** to pick up your changes. See [Test Scripts](/platform/test-scripts).
  </Accordion>
</AccordionGroup>

***

## Related guides

<CardGroup cols={2}>
  <Card title="Scenarios & Categories" icon="sitemap" href="/platform/scenarios-and-categories">
    Test cases live inside scenarios — start there if you haven't generated scenarios yet.
  </Card>

  <Card title="Test Scripts" icon="scroll-text" href="/platform/test-scripts">
    Next step: turn approved test cases into step-by-step scripts.
  </Card>

  <Card title="Workbooks" icon="book-open" href="/platform/workbooks">
    Test data comes from your workbook — keep it clean and the AI generates better cases.
  </Card>

  <Card title="Knowledge Wiki" icon="library" href="/knowledge-ai/knowledge-wiki">
    Add domain rules so the AI uses your real test data, not placeholders.
  </Card>
</CardGroup>