Skip to content

Organizations and billing

An organization holds your projects, your servers and your billing. The first time you sign in, Ployz creates one for you, named after you (like Ada's Projects). An agent can also make one before anyone signs in, named Agent sandbox. Open its claim link within 24 hours to keep it. See Try Ployz without an account.

You can invite people to work in your organization with you. See Invite people. To let a CI job act for you, make an organization token under Organization → Tokens. Pick Deploy and an Environment for a token that can only deploy there, Read only for one that can't change anything, or Full as you for one that can do whatever you can. Deploy needs an Environment, so in an organization without one the dialog starts at Read only.

To make another organization, open the account menu and click Switch organization → New organization. Name it, change its URL if you like, and click Create. You land in the new organization, and you own it. Switch between your organizations under Switch organization.

From the CLI, run ployz org new with a name:

ployz org new "Acme Corp"
ployz org new "Acme Corp" --slug acme

The slug comes from --slug, or from the name when you leave it out. If another organization already has that slug, Ployz adds a short suffix, and the command prints the slug it picked. The CLI then acts in the new organization. Your browser and your other signed-in CLIs stay where they were. An organization token can't make an organization, so sign in with ployz login first.

In the sidebar, click Organization for its settings:

  • General: your organization's Display name. Owners and admins can change it: type a new one and press Enter or click the check. Its URL stays the same. Below it, Allow support access, and Forget all servers for when your servers were deleted.
  • Network: your Cluster name, which your services' long private addresses carry. See Name your cluster. Below it, your organization's Domain, which your services' free addresses live under. Until you generate your first domain, it reads No domain yet.
  • Members: who's in the organization, their roles, and, for owners and admins, pending invites. See Invite people.
  • Tokens: organization tokens. See Make an organization token.
  • GitHub: the connected GitHub App. See Deploy from GitHub.
  • Builds: where builds run. See Where builds run.
  • Billing (when billing is on): your plan.

Organization → General: the domain and the Danger section

Everyone in an organization has one role: Owner, Admin or Member. Each role can do everything the role below it can, and more:

RoleCan do
MemberSee the organization, its members, projects, servers, volumes, variables, logs and metrics. Create projects and environments, edit projects, stage changes, deploy, destroy and merge them, restart and stop services, answer approvals, and make their own organization tokens.
AdminAlso change the organization's settings and support access, invite people, change roles and remove members, manage everyone's tokens and agents, connect GitHub, delete projects and environments, protect and unprotect environments, delete volume copies, and add, operate, open a shell on and remove servers.
OwnerAlso manage billing, delete the organization, and make or remove an owner.

When you try something your role can't do, Ployz says who can: "Only an owner or admin of this Organization can do this." or "Only an owner of this Organization can do this." A change to a project's configuration that your role can't make is refused with the permission it needs, such as Needs environment:delete; ployz <command> --help names that permission before you run it.

To invite someone, go to Organization → Members and click Invite. Pick Member or Admin, add their email if you like, and click Make link. When Ployz Cloud has email on, it emails the link to that address, and the dialog tells you where it went. A self-hosted Cloud without email, or an invite with no email, sends nothing, so copy the link from the dialog and send it to them yourself. Whoever opens the link, signs in and clicks Join and the organization's name joins with that role, so share it only with the person you mean. A link works once and expires after 7 days. Someone already in the organization who opens it sees You're in and Open instead, and keeps their role. Owners and admins see pending invites, who sent them and when they expire. Click Revoke to cancel one. The list updates as soon as someone joins.

Owners and admins change a role with the select beside each person, and click Remove to take someone out. To leave an organization yourself, click Leave on your own row. An organization always keeps an owner, so its last owner can't leave or step down until they make someone else an owner. ployz org member rm refuses to remove the last owner, and ployz org member role refuses to give them another role. Each refuses at once, before it asks for any approval.

When someone leaves or is removed, they lose the organization at once: its dashboard, their organization tokens, and their signed-in CLIs' access to its servers. If they joined by invite and open it again, it says their membership ended.

Turn on Allow support access under Organization → General to let Ployz support open your organization and do what you can. It's off until an owner or admin turns it on, and turning it off applies on support's next request. While support has your organization open, a banner at the top says so. A self-hosted Ployz Cloud has no Ployz support, so the setting doesn't appear.

Support sees your projects, services and settings. Three switches under the setting decide what more it sees:

  • Variables: values of variables that aren't sealed. Off, they read as sealed.
  • Config files: what's in each config file. Off, support sees each file's name and mount but not its contents.
  • Container logs: what your containers print, including a failed health check's output. What a build prints is always visible.

All three start off, and turning support access off turns them off. Each applies on support's next request. Support sees these switches but can't change them. It can still change anything else you can, including a value it can't read, but it can't delete the organization.

A coding agent working through the CLI can save or deploy a change that removes a running service, deletes a volume and its data, detaches a volume from a service, or removes a domain. It can also drain or remove a server, rename the cluster, or clean up what a deleted environment left on your servers. It can remove a project, an environment or the whole organization, forget every server, or turn an environment's protection off. Ployz refuses that command with approval_required, naming what it would remove and the approval it opened for you. Every organization works this way, and no setting turns it off. The one exception is an Agent sandbox nobody has claimed yet: no person could answer there, so nothing asks until someone keeps it.

How you answer depends on where the command runs:

  • In a terminal, it lists what it would remove and asks you to type the environment's name. A server command asks for the server's name, or for server clean the name you passed it. ployz org rm, and ployz server forget with no server named, ask for the organization's slug. Type it and the command carries on. Anything else, or Ctrl-C, leaves the approval pending and changes nothing.

  • Without a terminal, such as in a script, it prints the same list, the approval's ID, a link to the Ployz agent where it waits, and ployz org approval approve ID. With PLOYZ_TOKEN, when Cloud doesn't name the organization, the link is Ployz Cloud itself. A redeploy's list names the deploy it ships again, such as This redeploy of #4:, the later deploys it rolls back, and what it removes again that its deploy removed, such as ✕ removes Service extra. When a Protected environment asks, it prints Ployz Cloud's reason, that an owner or admin who didn't ask approves it in Ployz Cloud or with ployz org approval approve ID, and ployz org approval show ID to follow it, then waits for them. It waits until someone approves or denies it in Ployz Cloud, then carries on, or fails naming what was denied and the reason given. Ctrl-C, or losing Ployz Cloud, stops waiting and leaves the approval pending, and the error names the command to run again with --approval ID, except for ployz env sync --value ROW (see --json below), whose rerun asks again. If Ployz Cloud answers with an error instead, the command fails saying it couldn't check the approval, and names the command to run again only when the error is on Cloud's side. If Cloud refuses your sign-in or token, or you've left the organization, it names no command to run again; it says what to do instead, such as ployz login, a new token or ployz logout. Once you've left, ployz logout forgets the sign-in here though Cloud refuses to sign it out, and warns that Cloud didn't revoke it.

  • With --json, it never asks or waits. It exits with approval_required, and details.retry is the same command with --approval ID. Run that once the approval is approved. The one exception is ployz env sync --value ROW, which refuses --approval, since an approval can't carry secrets given again on stdin. Its retry names each --value ROW but never the secret and no approval, so pipe the values on stdin again and it asks for a new approval. To finish such a sync, run it without --json so it waits for the approval.

  • In an agent that uses ployz mcp, the agent's app asks you. In an app that can't ask, the call fails with a link, as for a connected agent below. See Give the agent Ployz tools.

  • In the Ployz agent in the dashboard, you approve or deny right in its panel, and the agent waits for you.

  • In an agent you connected to Ployz Cloud, the call fails with a link. Approve or deny under Waiting on you in the Ployz agent, then have the agent try again.

  • ployz deploy and ployz deploy --skip-deploy ask only when the change removes something that's running. Rolling out a new image or changing variables never asks.

  • Every deploy or redeploy an agent asks for asks, even one that destroys nothing. That covers the Ployz agent, ployz mcp and an agent you connected to Ployz Cloud. An agent's save still asks only when it removes something, and one approval covers one deploy or redeploy. Outside a Protected environment, redeploying from the CLI or the dashboard never asks, except from the CLI when it takes something off: redeploying a ployz env shutdown, or rolling back to a deploy from before a service was added, asks as a deploy that removes the same would. In a Protected environment, a member's save, deploy or redeploy asks an admin even when it removes nothing. An owner's or admin's Deploy token there redeploys unasked when the redeploy destroys nothing. A rollback, or a redeploy that removes again what its deploy removed, asks an admin. A rollback that would remove more, or a redeploy that would prune a service or can't check every server for one, is refused, naming what and that an owner or admin can redeploy it.

  • ployz project rm, ployz env rm and ployz env shutdown ask when they would take something off the servers. ployz org rm, ployz server forget with no server named and ployz env protect --off always ask.

  • Removing or forgetting a server Ployz Cloud manages with ployz server rm or ployz server forget SERVER always asks. ployz server drain asks when a global service would run on no other server, and ployz server clean asks when it would stop a service or delete volume data. ployz server upgrade never asks for an approval.

  • An agent's stop or restart of a service always asks, and one approval covers one stop or restart. Typed in a terminal, ployz service stop and ployz service restart don't ask yet. Starting a service never asks.

  • Dashboard buttons and dialogs never ask: confirming a dialog is your approval. The exception is a member's request in a Protected environment, such as a save, deploy, sync, restart, stop or volume grow, which waits for an owner or admin (see Permissions). When the Ployz agent does the same thing for you, it asks.

  • If the change moves on after Ployz asked, your approval no longer fits it, and the command asks again about the new one.

  • An approval covers one save. A later save that still carries a removal that hasn't deployed yet asks again.

A retry of a command you denied is still denied, even once it would destroy nothing. A retry of one still waiting for your answer still waits, unless it no longer destroys anything.

An approval works only for the person who asked for it, from the same login or token they asked from, and only for the command and the target they asked about. Anyone else who sends it is refused with forbidden, and nothing changes. So is the same person from another login or token, or a different command or target, even one that wouldn't ask for an approval at all, such as a save into another environment that removes nothing, a quick ployz volume sync, or one already done, such as ployz env protect --off for an environment that isn't protected. The refusal says which. One exception: when an agent retries a tool that finds nothing left to write, such as a save that already landed, only the person and the login or token are checked. Run the command again without --approval to ask for an approval of your own. When two people, or one person from two logins, ask for the same thing, each gets an approval of their own to answer.

An approval runs its command once. Sending it again for another run asks again, with a new approval to answer, even when the command and target are the same, such as changing a member back to a role you changed once already. A retry of the same deploy, or the same request sent again with its request key, answers what the first one started and runs nothing new. A save or deploy that Ployz refuses leaves its approval unused.

You can answer from a terminal too. ployz org approval ls lists the approvals waiting, ployz org approval show ID shows what one would remove, and ployz org approval approve ID or ployz org approval deny ID --reason "keep the volume" answers it. Approving asks you to type the organization's slug, so it needs a terminal.

An agent can't answer an approval: ployz mcp, an agent you connected to Ployz Cloud and the Ployz agent can neither approve nor deny one. When the Ployz agent or an agent you connected to Ployz Cloud asked, you approve in the Ployz agent's sidebar, and ployz org approval approve refuses with forbidden. When an agent using ployz mcp asked, you approve in its app's dialog, in the sidebar, or with ployz org approval approve. The dialog and the command answer with this computer's ployz login, so it must be signed in to the agent's Ployz Cloud and acting in the organization that owns the request. Otherwise they refuse: run ployz org use ORGANIZATION to switch, then answer again. If you run the same change yourself with ployz, typing the name in your terminal approves it. Anyone in the organization can deny any request, from the dashboard or a terminal.

An agent's own sandbox doesn't ask. While only the agent owns it, nobody could answer, so its deploys and removals go ahead. Once a person keeps it, the agent asks like anywhere else.

This keeps a well-behaved agent from deleting something by mistake. It doesn't stop a program that's trying to get around it: anything that holds your own ployz login or one of your own tokens can approve its own ployz request, as if you'd typed the name in a terminal. The agent token ployz mcp login saves can't approve or deny anything. Keep tokens you give an agent short-lived: a token can't make or revoke tokens, so revoking it cuts the agent off. Keep backups of data you can't lose.

Ployz is free on your own servers: unlimited servers and projects, preview environments, generated https addresses and custom domains. Your hosting provider bills you for the servers.

Pay when your cluster is in production. Production is $10 a month with two servers included, then $5 for each extra server. Production pays for what Ployz Cloud runs for you:

  • Uptime checks from outside your servers, and an email when a service goes down or comes back.
  • Web analytics for your domains.
  • Longer log and metrics history, coming.

Without Production, Ployz keeps checking your services and counting your visits. You can't see the results: the Monitoring tab, Analytics, ployz uptime and the Ployz agent ask you to add Production instead, and the uptime emails stop. Add Production and the results show again, as far back as Ployz keeps them. Everything else is free: metrics, logs, deploys and the crash emails.

Click Upgrade to production on Organization → Billing to start. The page shows your server count and what you pay a month. A server you add is charged as soon as it joins. When you remove one, the lower price starts at the next renewal. Manage billing on the same page cancels Production or gets invoices. A cancelled Production stays until the end of the period you paid for.

In an agent's sandbox, billing waits until someone keeps it.

A self-hosted Ployz Cloud has no billing. Custom domains, uptime and analytics are always available there.

Deleting an organization removes it with its settings and tokens, and disconnects your servers from it. It can't be undone. Only an owner can delete it. For anyone else Delete organization is disabled.

  1. Delete every project in it. Delete organization stays disabled until you do.
  2. Cancel Production first if you pay for it. Deleting the organization doesn't cancel it: go to Organization → Billing → Manage billing.
  3. Go to Organization → General and click Delete organization.
  4. Type the organization's slug, as the dialog shows it, and click Delete.

If a server is offline, the organization stays, disabled, until that server is back. Then click Try again. A server that now belongs to another cluster keeps the organization waiting the same way, since Ployz Cloud won't change it. Stop Ployz on that server, over SSH, with sudo systemctl stop ployz.socket ployz. Then forget your servers and click Try again. That server still belongs to the other cluster, so skip the uninstall step the Forget dialog mentions: start Ployz on it again with sudo systemctl start ployz.socket ployz once the forget finishes.

The dashboard then opens another of your organizations. With none left, it offers Create an organization.

Once it's gone, its slug is free, and a new organization can take it. Old links and emails that name the slug open the new organization only for its members, and for Ployz support if it allows support access. Once its Domain has served your services, no organization gets it again.

From the CLI, ployz org rm ORGANIZATION does the same. If this device was signed in to that organization, it moves to another of yours and says which. With none left, run ployz logout, then ployz login.

An agent's sandbox ends 24 hours after the agent made it, unless someone keeps it. Ployz then removes it. See When nobody claims it.