Playwright for pages you own

Playwright belongs on properties you own. Your shop, your docs, your staging host, a checkout you can break without paging a stranger. MaskWright does not ship Playwright. This how-to keeps that driver in a scripts folder, pointed at your hosts, outside the profile store.

The automation cluster already split browser automation versus a real profile and Playwright versus Puppeteer versus Selenium. This page is the scraping-desk application: when a headed driver is collection against furniture you control.

What "you own" means

A hostname you operate. A staging URL on your laptop. A production shop whose robots and terms you also write. A login you created for your own admin.

It does not mean a SaaS inbox you pay for. Paying for LinkedIn or Shopify does not make their HTML your property. Scraping behind a login is not research. Official APIs and exports are still the door on those hosts.

It does not mean a public Ads Library card you want to harvest. Facebook Ads Library research locally is a person or an official tool. Not Playwright.

Where the driver lives

A repo on disk, not under %APPDATA% next to maskwright-data. Disposable profile folders Playwright creates for a run. Logs you delete. Local scripts stay outside work rooms.

Do not point userDataDir at a MaskWright profile. Do not attach to a remote debugging port on a work profile. Internal CDP in 0.1 is for fingerprint and cookie apply. It is not a user automation API. What a scraping browser is is a collector product. MaskWright is not one.

Traces and HAR files from your own host can live in the research folder. Traces from someone else's login cannot. Storing research files.

Disposable directories

Playwright will create a profile folder if you ask it to. That folder is a fixture. Delete it after the run. Do not reuse it as a MaskWright profile. Do not copy cookies from maskwright-data into it so a spec can "be" the store login.

On Windows the fixture can live under the repo, for example E:\work\shop-tests\.pw-user. The store profile stays under the Electron user-data path. If you cannot point at two different trees, you have already mixed the jobs.

A headed run on localhost is still a driver. You will see automation flags. That is fine on a host you own. You are not hiding from yourself. You are checking a page.

What I will not turn this how-to into

A stealth launch recipe. We will not hide a scraper. No navigator.webdriver theater for third-party sites.

An anti-bot walkthrough. If your own site trips a wall you bought, you fix the allowlist. Anti-bot pages and official channels. You do not learn bypass on a stranger's wall.

A personal-data harvest against a store you do not operate. Personal data and collection.

A "Playwright plus antidetect" stack. Collection and operator profiles stay apart. A profile is a folder. A driver is a process.

Honest uses on a Windows desk

Screenshot your own marketing pages in a few viewports. Click your own checkout after a deploy. Pull text from your own docs into a fixture. Hit your own staging API through a real browser when a unit test is not enough.

Use an honest UA on your own host if you want. On someone else's host, this how-to does not apply.

A Friday deploy is a good test of the split. The spec hits https://staging.yourshop.test. The Seller Central profile stays closed. If the spec fails, you fix your page. You do not point the same spec at a competitor checkout to "see how they do it." That is someone else's host.

MaskWright 0.1 has no Playwright, Puppeteer, or Selenium integration. It will not grow one so a spec file can share cookies with Seller Central. If you need the driver, you already know the install command. Point it at a hostname you can defend.

The Scraping hub will keep this URL narrow. Own host, outside the store, no disguise. Everyone else's pages go back to official channels or a slow public look.

If a teammate checks in a spec that hits a third-party login, reject the PR. The spec is not coverage. It is a harvest with a test runner's name on it.

FAQ

Can I point Playwright at a MaskWright user-data-dir?

Do not. Keep the fixture under the repo. Keep work profiles in the app.

Does "I pay for Shopify" mean I own Shopify's HTML?

No. You own your shop's storefront if you operate it. You do not own the admin HTML as a scrape target. Use Shopify's APIs and exports for machine work.