Automation ethics we follow

This is the policy page for the automation cluster. It is not legal advice. It is the list I use when a keyword wants a bypass and the useful article is a boundary instead.

MaskWright 0.1 launches and isolates profiles on Windows. It is not an RPA host and not a Playwright product. The ethics have to match that machine.

Authorized work only

I write for operators who already have a right to the accounts they open. Brand pages they own. Ads accounts they were hired to run. Stores they registered. Research on public pages the terms allow. A contractor with a Business Manager role. A creator with a Brand Account. A seller with an entity they can name on a tax form.

I do not write create-unlimited guides. I do not write restore-after-ban steps.

If you need the architectural split, it is browser automation versus a real profile. Profiles for people. Drivers for properties you own.

Official channels first

If the platform published an API, a feed, or an export, that is the door. A selector on the production UI is a worse door.

If the platform published an official scheduler or business suite, use that before an unofficial helper.

If the platform said no to automation, we stop. We do not hide a driver. We do not buy a captcha solution. Captcha solvers are not research. We do not install a timing pack. Keyboard timing packs are theater.

Scripts stay out of work profiles

A live cookie store is not a test fixture. Local automation lives in a scripts directory, pointed at hosts you control, using disposable profile folders. Local scripts stay outside work rooms.

Do not leave a debugging port on a work profile for a helper to attach. Remote debugging ports and operator risk.

Do not treat an MCP path or an agent runtime as a local profile. MCP is not a local profile. Agent browsers versus operator profiles.

Batch is not a fleet

Opening several authorized profiles is operator convenience. Replaying one body across those profiles is a different product. Batch launch is not a farm. We did not ship the replay.

A stop is not a puzzle. The scraping cluster writes the collection version of this. This page is the automation version.

Local Windows notes

This cluster writes for a PC you can unplug. Profiles stay on the operator disk. There is no team-seat control plane that can replay your clicks from a browser in another country. That absence is ethical as well as technical. A control plane that can click every profile is a fleet waiting for a password.

Geo lookups, ads feeds, and the updater still leave the machine. We do not pretend the app is air-gapped. We pretend less than a cloud locker that holds the cookies.

If a contractor needs access, they get a platform role, not a mailed user-data folder. If a script needs access, it gets a key, not a work cookie.

What we will still help with

Choosing an official driver for tests you are allowed to run. Keeping research profiles empty of work cookies. Saying no to cloud runtimes that hold the session. Explaining why 0.1 has no agent, no RPA, and no stealth export.

If a request needs us to cross a line above, we skip the how-to, not just the feature. A competitor title can still be a keyword we rank for. The body will be the boundary, not a softer version of their guide.

The rest of the notes live under Automation. Read them as applications of this page, not as a toolkit that forgets it.

FAQ

Is this legal advice?

No. It is how we write and how 0.1 is built. Your counsel and the platform contract still win.

Can I automate accounts I own with MaskWright?

You can open those accounts in isolated profiles and work as a person. 0.1 will not replay clicks across them. For machine work, use the official API or a driver on a host you operate, outside the app.