Create test cases and test plans using Thunders AI.
A test case covers one behavior. A test plan groups related scenarios for a feature or workflow.
Thunders AI can create web, mobile, and API tests, and can mix App types in a single test case, for example web and API apps, or mobile and API apps.
Test cases
From a description
Describe one behavior and the expected outcome. Keep the prompt focused so the resulting test case has a clear purpose.
Example
You: Verify that a Funding Officer can sign in and create a new funding contract from the New Contract form. The contract should appear in the contracts list with a "Contract created successfully" confirmation.
From a live URL (Web Browse)
When the Web Browse source is enabled, you can include a URL and Thunders AI will open the page in a real browser to understand its structure: reading page titles, field labels, button text, and layout before writing the test.
Example
You: Go to https://prod-autofinance.lovable.app and write a test that adds a new dealer through the Add Dealer form and checks that the new dealer appears in the dealers table.
Use URLs when the page structure matters or when you want the test to start from a specific location. Web Browse inspects the live page so Thunders AI can reference exact element labels and states.
Note: Enabling Web Browse allows Thunders AI to interact with the target page. This may change page state, trigger form submissions, or navigate away. Use with caution on production environments.
From a ticket URL or ID
When a tool is connected and enabled, Thunders AI can use tickets as context for a test case. Provide either:
a ticket ID / key (e.g.
APP-477,KAN-6), ora ticket URL (full link).
Use one or the other. You don’t need to paste both.
Supported sources:
Jira: e.g. "Create a test case from KAN-6 in Jira"
Jira Data Center: same as Jira, for on-premise Jira instances
Xray Cloud / Xray Data Center: e.g. "Import test XSP-2 from Xray"
Confluence / Confluence Data Center: use spec pages as context
Linear: e.g. "Create a test case from APP-477 in Linear"
Tip: Combine a ticket with Web Browse for more reliable steps.
Example prompt: Read KAN-28 in Jira and create a test case. Once you have read the ticket, use Web Browse to validate that the steps you plan to author are reliable.
From an upload (documents & files)
Sometimes the test you want lives in a document or file. This can be a spec, PDF, CSV of test data, a screenshot, or even code you want to import (for example, a Playwright test file). Thunders AI can read uploads and use them as context for the test steps it generates.
Drop files into the conversation: upload files directly into a Thunders AI conversation from the input area.
Thunders AI reads the content: it extracts what’s inside and uses it as context for the generated steps.
Hand Thunders AI a feature spec, a Figma export, or a list of test data, and it builds tests that reflect what is in the file, without you having to paste long text into the chat.
From an existing test case
Thunders AI can search your project's existing test cases and reuse them as steps inside a new test. This is useful for common flows like login.
Example
You: Create a test that starts with the login flow and then navigates to the settings page to update the user's display name.
Thunders AI searches for an existing login test case in your project and adds it as a reusable step, then generates the remaining steps.
Using environment variables
If your project has environments configured with variables, Thunders AI uses them automatically. For example, if your environment has APP_AUTOFINANCE_URL, Thunders AI will use [APP_AUTOFINANCE_URL] in the navigation step instead of a hardcoded URL.
You can also reference variables explicitly in your prompt:
Example
You: Navigate to [APP_AUTOFINANCE_URL], sign in with [FUNDING_USER_EMAIL] and [PASSWORD], and create a new funding contract.
API tests
For API tests, describe the request, authentication, and expected response.
Example
You: (with your API documentation attached) Use the attached doc to author my test: call the stats endpoint and capture contracts_total, then add the Login test case, go to the contracts dashboard, and assert the contract count shown equals contracts_total.
Thunders AI also supports mechanical assertions and extractions for API responses: for example, capturing a value from one API call into a variable and using it in a subsequent step. You can also mix API and AI steps in one test:
Step types Thunders AI can generate
Thunders AI selects the right step type automatically based on your description:
Step type | Used for |
AI (default) | Clicks, typing, navigation, assertions, and variable operations. Most steps use this type. |
API | Explicit API call requests (GET, POST, etc.). |
Web | Low-level web browser actions: refresh, go back/forward, switch tabs, wait for page load. |
TestCase | Referencing an existing test case as a reusable step. |
Conditional | Branching logic: go to a different step based on a condition. |
VisualCompare | Comparing a page screenshot against a baseline image asset. |
FileCompare | Comparing two file assets (e.g. a downloaded file vs. a baseline) at the byte or structural level. |
Test plans
A test plan groups related scenarios for a feature or workflow. Describe the feature, attach a document, or use a connected source, and Thunders AI generates a structured plan with test sections, scenarios, and draft test cases you can promote into your project.
Example
You: As a registered user, I want to log in to my account using my email and password so that I can securely access my dashboard, while being protected against unauthorized access and common attack vectors. Use [FUNDING_USER_EMAIL] and [PASSWORD].
For the full guide, including how to reuse your project and environment variables in the generated plan, see Generate Test Plans with Existing Variables.