# Permissions

> Who can do what in Ployz. Roles, tokens, Protected environments and approvals in one place.

Three things decide what you can do in Ployz: your role, the token you use, and whether the
environment is Protected. You get the smallest of the three. A token never does more than your
role, and a Protected environment holds some changes back even from a token that could deploy.

## Pick a role for each person

Everyone in an organization is an **Owner**, **Admin** or **Member**. Members deploy, stage
changes and answer approvals. Admins also invite people, manage servers and delete projects.
Owners also manage billing. The full table is under
[Invite people](https://ployz.dev/docs/account/organizations#invite-people).

When your role can't do something, Ployz says who can, such as "Only an owner or admin of this
Organization can do this."

## Narrow a token to one job

A token acts as you, narrowed by its preset:

- **Full** does everything you can.
- **Read only** reads and changes nothing.
- **Deploy** reads what you can read, but stages and deploys changes only in the one environment
  you name.

Give CI a Deploy token for the one environment it ships to. See
[Make an organization token](https://ployz.dev/docs/cli/overview#make-an-organization-token). An agent you connect
with `ployz mcp login` gets its own token, and that token can't answer approvals.

## Protect production

A Protected environment keeps its riskiest changes for owners and admins. Only they can connect a
repository, change the deploy branch, turn protection off, or make a Deploy token for it. A new
project's default environment starts Protected. See
[Protect an environment](https://ployz.dev/docs/environments/environments#protect-an-environment).

A member can still ask. Their **Deploy** button reads **Request deploy**, and the request waits for
an owner or admin who didn't ask.

A merge into a Protected environment's deploy branch still deploys. Use GitHub branch protection to
decide who can merge. See [Protected environments](https://ployz.dev/docs/deploy/github#protected-environments).

## Answer approvals

Some changes wait for a person before they run:

- every deploy an agent starts;
- anything an agent or the CLI does that would delete a service, volume or domain;
- a member's request in a Protected environment.

A delete from the CLI asks in your terminal first. See
[Destructive actions ask first](https://ployz.dev/docs/account/organizations#destructive-actions-ask-first). The rest wait under **Waiting on you** in the [Ployz agent](https://ployz.dev/docs/account/agent-sidebar#approve-or-deny-a-deploy).
Read the card, then click **Approve** or **Deny**. An agent you connected to Ployz Cloud waits the
same way. See [Approve a deploy](https://ployz.dev/docs/account/connect-an-agent#approve-a-deploy).

## Good to know

- Nobody approves their own request, and an agent can't approve its own deploy.
- Protected guards what runs through Ployz. An admin with a shell on a server can still reach any
  environment on it.
- Leaving an organization ends your access and your tokens at once.

**Next:** [protect your production environment](https://ployz.dev/docs/environments/environments#protect-an-environment).