Examples

How to read these

Each example shows the prompt you type and the tools INGenie (or any connected MCP client) calls on your behalf. You never invoke tools by hand — but knowing which ones exist makes your prompts much sharper.

1. Author a browser test from a description

You: Create a test under scenario Login called LoginSuccessful for https://practice.expandtesting.com/login. Sign in with practice / SuperSecretPassword! and assert the success banner.

#	Tool	Effect
1	`ingenious_browser_discover`	Opens a real browser and walks the flow.
2	`ingenious_browser_session_save`	Writes stable locators to `ObjectRepository/LoginPage.yaml`.
3	`ingenious_testcase_create`	Creates `TestPlan/Login/LoginSuccessful.csv` with linked steps.
4	`ingenious_testcase_validate`	Checks the result against the authoring conventions.
5	`ingenious_run`	Executes it and returns the report.

Give it the URL

Discovery is far more reliable than guesswork. If you supply a reachable URL, INGenie derives locators from the live DOM instead of inventing them.

2. Convert manual test steps

You: Convert these manual steps into a test case under Checkout / GuestCheckout:

1. Open the shop homepage
2. Add the first product to the basket
3. Continue as guest
4. Fill in the delivery address
5. Verify the order confirmation number is displayed

Uses the convert_manual_steps prompt → ingenious_action_search (to map each sentence to a real action) → ingenious_testcase_create → ingenious_testcase_validate.

3. Make a test data-driven

You: Parameterize Login/LoginSuccessful so it runs for admin, standard and locked-out users.

#	Tool	Effect
1	`ingenious_testcase_parameterize` (mode `scan`)	Proposes which literals should become columns.
2	`ingenious_data_sheet_create`	Creates the data sheet.
3	`ingenious_testcase_parameterize` (apply)	Rewrites the steps to `Sheet:Column` references.
4	`ingenious_data_row_add`	Adds one row per user type.

Review the scan first

Run the scan mode, read the proposal, then apply. Blanket parameterization tends to externalise values that are genuinely constant.

4. Build an API test from a cURL command

You: Import this cURL into scenario Users and turn it into a test case that asserts HTTP 200 and that email is present.

curl -X GET "https://api.example.com/users/42" -H "Authorization: Bearer $TOKEN"

→ ingenious_import_curl → ingenious_testcase_create → ingenious_testcase_validate → ingenious_run.

The same flow works with ingenious_import_postman, ingenious_import_bruno, ingenious_gen_from_openapi and ingenious_gen_from_har.

Secrets stay out of the repo

Replace tokens with an environment or Global Data reference before you let INGenie create the test case. A cURL pasted verbatim will otherwise be committed with the bearer token inside it.

5. Run a suite and explain the failures

You: Run the Regression test set on Chromium and tell me why things failed.

#	Tool	Effect
1	`ingenious_run_async`	Starts the run, returns a `runId`.
2	`ingenious_run_status`	Polled until the run completes.
3	`ingenious_report_failures`	Ranks the top failure reasons.
4	`ingenious_report_compare`	Diffs against the previous run to separate regressions from known failures.

6. Harden a flaky test

You: Checkout/GuestCheckout is flaky — harden it.

Uses the harden_test prompt: ingenious_testcase_show → ingenious_object_show → replaces index-based locators, removes fixed sleeps in favour of explicit waits, adds assertions → ingenious_testcase_validate.

7. Turn a functional test into a load test

You: Build a k6 load test from Login/LoginSuccessful with a spike profile.

#	Tool	Effect
1	`ingenious_perf_export`	Generates the k6 script.
2	`ingenious_perf_validate`	Dry-runs with 1 VU / 1 iteration.
3	`ingenious_perf_run_async`	Starts the load test, returns a dashboard URL.
4	`ingenious_perf_status`	Live VUs, RPS, p95 and error rate.
5	`ingenious_perf_compare`	Regression check against a baseline.

Load testing is not a dry run

ingenious_perf_run generates real traffic against a real system. Confirm the target is a test environment, and that you are allowed to load it, before approving the tool call.

8. Extract reusable components

You: Look across the Checkout scenario and pull repeated step sequences into reusable components.

Uses the extract_reusable prompt → ingenious_testcase_list / _show → creates reusable scenarios → rewrites the callers with Execute steps.

9. Health check

You: Is my setup complete?

→ ingenious_doctor reports the detected JDK, Playwright browsers, drivers and k6, and tells you exactly what is missing.

Next up :

Best Practices MCP Server