Skip to content
ployz
Deploy

Apps and databases on your servers, deployed by your coding agent.

Ployz turns the servers you rent into a platform your coding agent runs. It adds servers, ships the repo, sets up databases and opens a preview for every pull request. Deleting a service, a volume or a domain waits for you.

Connect your agentDeploy
your coding agent · ~/acme
  1. you › Read ployz.dev/agents.md and deploy this repo to my server.
  2. $ ployz server add root@203.0.113.12
  3. $ ployz service add web --repo acme/app
  4. agent › Staged on your canvas. Deploy?
  5. you › Yes.
  6. $ ployz deploy
  7. web: deployed · https://acme.com

ployz.dev · acme / production

web
acme.com
Online
Deployment #1 · applied

One pasted line, and your agent adds the server you rent and ships the repo.

Building blocks

One call for each thing you'd do over SSH.

Give an agent root and it has to get every step right by hand. Give it Ployz and each step is one call it can't half-finish.

  1. Add a server

    BeforeInstall Docker, generate WireGuard keys, add every peer to every server, open the ports, set up Caddy. Then do it again for the next box.

    Afterployz server add. The server joins the private network and takes services.

  2. Deploy

    BeforeWrite a Dockerfile, run a registry, build, stop the old container, start the new one, hope the health check passes.

    Afterployz deploy, or push the repo. No Dockerfile for eight languages, no registry, no downtime when a health check is set.

  3. Add a database

    BeforePick an image and a volume, generate a password, paste it into three .env files, write a backup script.

    Afterployz service add postgres. Wired to the app by name, with a sealed password and a restore point.

  4. Open a preview

    BeforeA CI job that copies the stack, a DNS record, a certificate, a database to seed, and a teardown nobody remembers.

    AfterEvery pull request gets its own https address. The merge ships it, then deletes it.

  5. Add a domain

    BeforePoint DNS, run certbot, renew it, write the proxy config.

    Afterployz domain add. The certificate is automatic.

  6. Read the context

    BeforeAn API with a tool for every endpoint, and the agent reads all of them first.

    After55 tools in ployz mcp, --json on every command, and ployz schema --json lists every command and setting in one call.

The guardrail

Your agent deploys. Deletes wait for you.

Paste one line into Claude Code, Codex or Cursor. It installs Ployz, adds your server and ships the repo. What can't be undone waits for you.

Read ployz.dev/agents.md and deploy this repo to my server.
your coding agent · ~/acme
  1. you › Postgres 17 is live. Remove postgres-14 and its volume.
  2. $ ployz service rm postgres-14
  3. $ ployz volume rm pgdata-14
  4. $ ployz deploy
  5. Waiting for your approval · remove service postgres-14, volume pgdata-14
  6. Approved by you in Ployz Cloud
  7. Deployment #19 of acme/production: applied postgres-14: removed pgdata-14: removed

ployz.dev · acme / production

postgres
Online
pgdata
postgres-14
Removed
Deployment #19 · applied

Stage everything. Ship once.

The agent's changes wait on the canvas until one Deploy ships them.

It stops before it deletes.

Deleting a service or a volume, detaching a volume, removing a domain or draining a server waits for a person. No setting turns it off.

Its token can't approve.

An agent connected through ployz mcp can't answer the approval. You approve from your terminal, Ployz Cloud or the dashboard's agent panel.

The parts you'd rent a platform for.

Deploy the whole app, not the parts your platform allows.

Web, workers, a queue, a self-hosted Sentry. Anything that runs in a container runs here, beside your app.

Push the repo. It builds.

Ployz detects the language and builds it. No Dockerfile, unless you want one.

Rent for the app, not for the build.

Your server builds by default. One workflow file hands the build to GitHub Actions, so a small server runs what a big one built.

Images run too.

Not everything you ship is your code. Point Ployz at an image, public or private, and it runs beside your app.

Private by default. Public on purpose.

Services reach each other by name across all your servers. The internet sees only your domains.

Networking that's live when the app is.

Private connections between your servers, SSL on every domain, nothing exposed by default. No VPN or firewall rules to write.

Postgres, Redis and the rest, one click.

Pick it from the canvas; it lands on the server you choose, wired to your app by name.

Variables that follow the environment.

Reference the database by name and every preview gets its own credentials, filled in for it. Nothing to copy between .env files.

Previews

See every change before it ships.

Every pull request gets its own link and its own database. Check what your agent built, then merge, and production gets exactly that.

Every pull request at its own link

Every push redeploys it. Anyone with the link can check it.

The merge ships what you saw

Production gets the code and the preview's setting changes, on a list you can edit. Then the preview and its data are deleted. Undo is in History.

Its own database

Pick which services a preview copies. A copied postgres belongs to the preview and starts empty. A setup command like pnpm db:seed fills it.

Shared only on purpose

A database you don't copy is shared with production, and the dashboard marks it in amber.

Credentials that follow

Variables that reference the database get each preview's own credentials.

One server to start. More when you need them.

No quorum to keep alive, no control plane to babysit. Every server is a peer, so losing one takes nothing else with it.

Add a server with one line

A VM or bare metal from any provider. A box at home behind NAT joins too; visitors come in through a server with a public address.

Move a service

Its database and files go with it, and the old server can go.

No server is the boss

So one box dying takes nothing else with it.

Up to 50 replicas

Pick the servers and the replica count for each service.

Everything after the deploy, included.

No uptime monitor, metrics or log service to sign up for and wire in. They're already in the dashboard, and your agent can read them too.

Monitoring

CPU, memory, disk and network for every service and server.

Searchable logs

Every service on every server, in one search.

Uptime checks

See when your app was down, and for how long.

Building for clients?

Each client on the servers it needs.

Give a big client its own box and keep small ones together. Send the preview link; they need no account.

A project per client

Unlimited projects, each on the servers its client needs.

Their data stays put

Every database and volume lives on a server you pick, yours or theirs.

Any provider

Any Linux box with systemd you can reach as root over SSH, a VM or bare metal.

Why your own servers

Yours, down to the engine.

Ployz is open source and runs on servers you rent.

Open source

The engine is Apache-2.0 and the dashboard AGPL-3.0. Host the dashboard yourself if you like.

Nothing refused

A self-hosted Sentry runs beside the app.

Change provider later

Add a server or leave one without starting over.

Pricing

Free

$0/ month

Everything to build and ship an app on one server.

  • One server, from any provider or a box at home
  • Unlimited projects, deploys and members
  • Push to deploy, and a preview for every pull request
  • One-click Postgres, MySQL, Redis and MongoDB
  • Custom domains with https
  • Monitoring, logs and uptime checks
  • ployz mcp for your coding agent, with approval before it removes anything
  • The dashboard agent, included for now
Get started

ProComing soon

$19/ month, 3 servers included

For when the project matters: users, revenue, data you can’t lose. $5 for each extra server.

  • Everything in Free
  • Backups off your servers, 30 days of restore pointscoming
  • A restore drill every week, with the result emailed to youcoming
  • Protected environments: your agent proposes, you approve
  • Roll back to any deploy from the last 30 dayscoming
  • Uptime checks from 4 regions
  • 90 days of metrics, analytics and agent activitycoming
  • Email support, next business day
Coming soon

ScaleComing soon

$999/ month and up, billed yearly

For a company that buys through its engineering lead.

  • Everything in Pro
  • SSO, SAML and SCIMcoming
  • Approval and token rules for the whole organizationcoming
  • An audit log of people and agents, with SIEM exportcoming
  • A commercial license for the dashboardcoming
  • A named engineer and a shared Slack channel
  • Onboarding and migration, done with you
Talk to us

Not per seat

Bring your whole team. Members are unlimited on every plan.

No surprise bills

Pro is $19 with 3 servers and $5 for each one after. Past an allowance, we warn you; we don’t bill you.

Leave with everything

Your servers, your data, and open-source parts you can SSH into.

Questions

Which coding agents work?

Claude Code, Codex, Cursor and any agent that runs commands or speaks MCP. Paste one line and it reads ployz.dev/agents.md, installs the CLI and connects ployz mcp.

What can my agent do without me?

Add servers, deploy, add databases and domains, open previews, read logs and metrics, and restart services. Its changes stage on the canvas until a Deploy ships them.

What waits for my approval?

Deleting a service or a volume, detaching a volume, removing a domain and draining a server. No setting turns it off. An agent connected through ployz mcp can't answer the approval; you approve from your terminal, Ployz Cloud or the dashboard's agent panel.

Why not give my agent SSH?

Over SSH every step is hand work the agent has to get right, and nothing stops it deleting a volume. Through Ployz each step is one call, and what can't be undone waits for you.

Which servers can I use?

Any Linux server you can reach as root over SSH, from any provider: a cloud VM, a bare-metal box, yours or a client's. A machine at home behind NAT joins too and runs services like the rest; it can't take web traffic, so keep one server with a public address and ports 80 and 443 open for visitors.

What happens when a server dies?

The others keep running. No server is the boss, so losing one takes down only what ran on it. Your dashboard still works, and you deploy those services to another box.

What will this cost me?

Nothing on one server. The Free plan covers one server with unlimited projects, deploys, previews, databases and custom domains. Pro is coming at $19 a month with 3 servers included and $5 for each extra server. Your provider bills you for the servers themselves.

Why would I pay for Pro?

Because the project matters now: it has users, revenue or data you can’t lose. Pro keeps 30 days of backups off your servers, where your agent can’t reach them, and test-restores them every week. It adds servers, longer history and email support. Pro is coming soon.

Does the dashboard agent cost extra?

Not for now. Ployz pays for it on every plan. Later it moves to pay per use, or you bring your own Anthropic key.

Can I leave?

Yes. Your servers, your data and standard parts you can SSH into. The engine is Apache-2.0 and the dashboard AGPL-3.0, and you can host the dashboard yourself.

Will my app work without a Dockerfile?

Node, Python, Ruby, PHP, Go, Rust, Java, Elixir and more build without one. Your own Dockerfile works too, and so does any Docker image.

Can the app run on my client's server?

Yes. Add their server with one line, on any provider, and deploy that client's services to it. Their database and volumes stay on that box.

The cheapest server is now the best place to ship.

Deploy