A production-ready Selenium framework. Python 3.11+. BasePage with 13 methods. POM pattern. pytest + pytest-xdist parallel. 5 custom ExpectedConditions. Retry decorators. Faker factories. Allure reporting. GitHub Actions matrix CI. 105 passing tests, with Ruff and strict production-code mypy checks clean. $995.
Already own a kit? Sign in for $895 owner pricing.
Twelve production-grade pieces of a Selenium test automation framework — wired together, documented, and ready to ship.
Drop in a Postgres, MySQL, or MongoDB connection string and pull random rows straight from your real DB — no mocks, no fixtures to maintain. Unicode + apostrophe + locale safety, seeded reproducibility, and an identifier-injection guard on every query. CI-verified against live databases every release.
413-line BasePage with find / find_visible / safe_click / type_text / hover / scroll_into_view / wait_for_url_contains / select_dropdown / screenshot — typed, logged, and retry-safe. Every page object inherits from it. Stop rewriting wait logic.
LoginPage, InventoryPage, CartPage, CheckoutPage — real data-test selectors, chaining returns, module-level Locator constants. The canonical Python POM pattern, done right.
A 346-line conftest.py with driver fixture, login fixture, auto-screenshot-on-failure hook, and Allure environment writer. --browser / --headless / --base-url CLI options out of the box. Same suite runs local and CI.
text_to_be_not_empty, text_to_contain, element_count_to_be, element_count_at_least, attribute_value_to_be — plus poll_until and wait_for_elements helpers. Replace every time.sleep() in your suite with a real condition.
StaleElementReferenceException is how Selenium works, not a bug to fight. The retry_on_exception decorator factory (and pre-configured retry_stale variant) re-runs the locator lookup, not the dead reference.
build_user(), build_product(), build_checkout() — Faker-powered, seedable for repro. Plus JSON-backed canonical fixtures (users.json, products.json) for tests that need exact values.
Every test annotated with @allure.story, @allure.severity, and with allure.step() contexts. Auto-screenshot on failure via pytest_runtest_makereport. Environment metadata written via pytest_sessionstart.
No chromedriver or geckodriver pinning. No webdriver-manager dependency. Selenium 4.6+ ships with Selenium Manager — drivers resolve automatically on every machine and every CI runner.
3 shards × 2 browsers (Chrome + Firefox) = 6 concurrent jobs. --reruns=2 for transient failures. Allure aggregation job uploads a single report artifact. E2E Complete status check for branch protection.
pyproject.toml, ruff.toml, mypy.ini all pre-configured. Type-hinted end to end. Lint workflow runs ruff + black + mypy in three parallel jobs under 60 seconds.
Login, inventory, checkout, and API-vs-UI cart specs run against saucedemo.com. Clone, pip install, pytest — you see a real test suite go green in minutes.
15 KB README plus 7 deep-dive guides in docs/: PAGE_OBJECT_PATTERN, WAITS_AND_FLAKE, FIXTURES_AND_FACTORIES, PARALLEL_EXECUTION, CI_CD, TROUBLESHOOTING. 2,500+ lines of explanation.
Real snippets from the kit. Type-hinted, Pythonic, retry-aware.
src/pages/login_page.pyfrom selenium.webdriver.common.by import By
from .base_page import BasePage
class LoginPage(BasePage):
USERNAME = (By.CSS_SELECTOR, "[data-test='username']")
PASSWORD = (By.CSS_SELECTOR, "[data-test='password']")
LOGIN_BTN = (By.CSS_SELECTOR, "[data-test='login-button']")
def login(self, username: str, password: str) -> "InventoryPage":
self.type_text(self.USERNAME, username)
self.type_text(self.PASSWORD, password)
self.safe_click(self.LOGIN_BTN)
return InventoryPage(self.driver)tests/e2e/test_checkout.pyimport allure
import pytest
from src.factories import build_checkout
@allure.story("Checkout")
@pytest.mark.e2e
def test_checkout_completes(logged_in_inventory):
checkout = build_checkout()
with allure.step("Add item and start checkout"):
cart = (logged_in_inventory
.add_to_cart("sauce-labs-backpack")
.go_to_cart())
confirmation = (cart
.checkout()
.fill_information(checkout)
.finish())
assert confirmation.is_complete().github/workflows/e2e.ymlname: E2E Tests
on: [push, pull_request]
jobs:
e2e:
runs-on: ubuntu-latest
strategy:
fail-fast: false
matrix:
browser: [chrome, firefox]
shard: [1, 2, 3]
steps:
- uses: actions/checkout@v4
- uses: actions/setup-python@v5
with: { python-version: '3.12' }
- run: pip install -e ".[dev]"
- run: pytest --browser=${{ matrix.browser }} --headless
env: { PYTEST_SHARD: ${{ matrix.shard }}/3 }Where Selenium Pro Python sits next to pip-installing Selenium yourself, other paid kits, and the Cypress / Playwright alternatives.
pip install gives you the WebDriver client and nothing else. This kit gives you the entire production framework AND the CI pipeline AND 105 passing UI, API, database, and framework tests. You start at week 4, not day 1.
Same scope as paid kits priced 2-5x higher. No subscription, no per-seat pricing, no vendor lock-in. One-time $995, lifetime access, instant download.
Pick Selenium if your team is already on Selenium, if you need the W3C WebDriver standard, or if your shop is Python-first and pytest is the test runner. Cypress Pro and Playwright Pro exist as sibling kits if your tooling points the other way.
Buy the kit when the reusable engineering foundation is the work you need to avoid. Keep building internally when the framework itself is your team’s product or learning goal.
If the kit avoids about 20 engineering hours at a $50/hour loaded cost, it has paid for itself. That threshold excludes future maintenance, design review, CI debugging, documentation, and onboarding time.
Use your own loaded labor cost and expected setup time; this is a decision aid, not a promised saving.Compatibility is stated against the current published package. Connect the kit to your own application, credentials, environments, and data after download.
Release metadata unavailableYour dashboard retains access to the version delivered with your purchase. Reasonable bug-fix and compatibility updates may be released, but no perpetual update cadence or operation of every future third-party dependency is promised.
Compare the profile above with your organization’s pinned runtime, browser, CI, proxy, and security requirements. For an unlisted constraint, ask before checkout rather than discovering it during setup.
Ask a compatibility question →Drop in a connection string and pull random rows straight from your Postgres, MySQL, or MongoDB. Unicode names, apostrophes, NULLs, weird locales — all the production data that breaks mock-based tests gets exercised every run.
tests/test_checkout.pyfrom framework.services.real_data import real_data
db = real_data(kind="postgres", url=os.environ["DB_URL"])
db.connect()
# Pull a real customer with a real order from your DB
customer = db.get_random_rows(
"customers",
count=1,
where={"tier": {"op": "eq", "value": "enterprise"}},
seed="checkout-fixture-1", # reproducible across runs
)[0]
driver.get(f"/account/{customer['id']}")
assert driver.find_element(By.CSS_SELECTOR, "[data-test=name]").text == customer["name"]
assert driver.find_element(By.CSS_SELECTOR, "[data-test=email]").text == customer["email"]
assert driver.find_element(By.CSS_SELECTOR, "[data-test=country]").text == customer["country"]Every release runs the kit's full test suite against ephemeral Postgres, MySQL, and MongoDB containers in GitHub Actions. The proof image just below this section is the actual terminal output of that run — no edits, no rebuilds.
Reproducible Python 3.12 + Selenium 4 validation against saucedemo.com. Current approved CI evidence is shown when available; otherwise the dated archival capture remains clearly labeled.
No. The kit ships with pure GitHub Actions parallel sharding via pytest-xdist — no Selenium Grid, Sauce Labs, or BrowserStack subscription required. The docs/CI_CD.md guide covers how to add a remote grid when you are ready.
Selenium 4.x. Selenium Manager (built into Selenium 4.6+) handles driver downloads automatically — no chromedriver/geckodriver pinning, ever.
The retry_on_exception decorator (and its pre-configured retry_stale variant) re-runs the locator lookup on StaleElementReferenceException, not the dead reference. The full pattern is documented in docs/WAITS_AND_FLAKE.md.
Yes. The framework is application-agnostic — point the --base-url CLI option at your local dev server and the same suite runs against it.
Yes. CI matrix runs on Ubuntu; the framework itself runs on all three OSes. macOS Apple Silicon supported.
Yes. The 15 KB README walks through every step, from pip install to your first passing test. docs/PAGE_OBJECT_PATTERN.md and docs/WAITS_AND_FLAKE.md then teach you the framework patterns so you can extend it.
Selenium 4 is the W3C WebDriver standard. Pick this kit if your team is already on Selenium, if you need cross-browser coverage across the full browser matrix (Safari, IE-equivalents), or if your shop is Python-first and pytest is the test runner.
Python 3.11+. Tested on 3.11 and 3.12. Uses modern type hints and pyproject.toml — no requirements.txt, no legacy setup.py.
Yes. Email support for setup questions and reproducible product defects is included with purchase. Response times vary with request volume; the contact form is also available for pre-sales questions.
An 11-page PDF walking through every directory, every pattern, and the exact flow from npx playwright test down to the system under test.