DevOps & Tooling · Gemini API and Google AI Studio
Playwright End-to-End Testing Gemini System Prompt
System instructions for Google AI Studio and the Gemini API, for Playwright End-to-End Testing work.
- Path
- prompts/gemini-system-instructions.json
- Loads
- Sent per request
- Format
- JSON system instruction
- Used by
- Gemini API and Google AI Studio
What this file tells the agent
A system instruction in Google AI Studio or the Gemini API gives the model the Playwright End-to-End Testing role, constraints and output format before every request.
- Strictly ban hardcoded delays (page.waitForTimeout()); rely exclusively on Playwright web-first auto-waiting assertions (expect(locator).toBeVisible()).
- Prioritize user-facing accessible locators: page.getByRole(), page.getByLabel(), and page.getByTestId(); never write fragile CSS selectors or brittle XPaths.
- Encapsulate multi-step interactions inside modular Page Object Models (POM) with typed parameters.
- Ensure total test isolation: authenticate via storage state (storageState) instead of repeating manual UI logins in every test.
- Configure trace viewer and video capture on first retry (trace: 'on-first-retry') for painless CI debugging.
Code to avoid, and what to write instead
The rule steers the agent away from the left and toward the right.
// Flaky: Brittle selectors, arbitrary sleep timeouts, and no auto-waiting
test("bad test", async ({ page }) => {
await page.goto("/catalog");
await page.click("div.col-md-4 > button:nth-child(2)"); // Breaks on minor CSS tweak!
await page.waitForTimeout(5000); // Flaky anti-pattern!
const text = await page.innerText("#cart-total");
expect(text).toBe("1");
});import { test, expect } from "@playwright/test";
test.describe("Checkout Flow", () => {
test("allows user to complete cart purchase", async ({ page }) => {
await page.goto("/catalog");
// Accessible, user-facing locators
await page.getByRole("button", { name: "Add to Cart" }).first().click();
await page.getByRole("link", { name: "Cart (1)" }).click();
// Auto-waiting assertions
await expect(page.getByRole("heading", { name: "Your Shopping Cart" })).toBeVisible();
await page.getByRole("button", { name: "Proceed to Checkout" }).click();
await expect(page).toHaveURL(/.*checkout/);
});
});Install it in one command
- 01
Go to your project root
The file goes at prompts/gemini-system-instructions.json, relative to the root.
- 02
Download the file
Run the command on the right. It downloads this preset and writes the file.
- 03
Make sure it gets used
Nothing loads this file automatically. Pass it as the system instruction in your Gemini API call, or paste it into Google AI Studio.
$ mkdir -p "prompts" && curl -fsSL "https://www.devscratchpad.tech/api/raw/gemini-prompts/playwright-e2e" -o "prompts/gemini-system-instructions.json"
writes prompts/gemini-system-instructions.json
prompts/gemini-system-instructions.json
To write the file from your own code instead, use Read your code or Convert Legacy Rules in the editor above.
More for Playwright End-to-End Testing
Questions
- How do I use this prompt in Google AI Studio?
- Paste the text parts into the System instructions field in Google AI Studio, or load the JSON in code and pass it as the system instruction (system_instruction in REST, config.systemInstruction in the Google Gen AI SDK).
- Which Gemini models support these instructions?
- Current Gemini models in the Gemini API and Google AI Studio accept a system instruction. Check Google's model list for the one you use.