Your coding agent can already create cloud infrastructure. Planton makes it verifiable, recorded, and reusable — in your account, on your laptop, free.
The whole Planton platform runs on your laptop and deploys to your cloud with the logins already on your machine. No account. Free for individuals, including commercial use.
You Already Have Two Ways to Do This
A Platform That Hides the Cloud
Convenience, in exchange for control.
Your app runs in someone else's account, under IAM you did not write, on a bill you cannot itemize. When you outgrow it, you start over.
Your Agent and the Cloud CLI
Control, in exchange for a record.
The agent writes the Terraform from memory and you learn at apply time what it got wrong. Permissions nobody derived. A cost you find out about next month. A database password that went through the chat to reach the shell. No pipeline. The second environment is a second conversation.
Planton is a platform too. It just runs on your machine, in your account, and gives the agent rails you can inspect.
For a one-off bucket, the agent alone wins. Planton earns its place on anything you will still be running in a month.
Two Ways to Ask. One Craft.
Ask from inside the tool you already use, or ask the assistant built in. Both run the same skills.
Your Coding Agent
Install the Planton skills once. From then on Cursor, Claude Code, or Codex composes infrastructure grounded in the real schema, validates it offline, and applies it through the platform on your laptop, while you stay in your editor.
npx skills add plantonhq/skillsWhat Planton Adds to the Agent You Already Use
A typed catalog, not memory: schema lookups and validation run offline, so a wrong field fails before it touches your cloud.
Verified before created: the monthly cost with its coverage stated, and the least-privilege permission policy derived from what is composed.
A record, not a transcript: every deploy is a stack job with a live log and a revision history; state sits under a path you can list.
Secrets the agent never reads: encrypted locally, key in your OS keychain, resolved on the runner at the moment of use.
Push-to-deploy from your laptop: it watches GitHub, builds in a pod, deploys to your cloud, and writes the status back.
What you built becomes a template: an Infra Chart redeploys into the next environment; agent commands do not compose.
When you become a team, nothing is redone: the same manifests and model on planton.ai or your own cluster.
The Built-In Assistant
Open the app and ask. The Planton Assistant converses with zero keys and zero accounts; Planton funds it.
Prefer to keep everything on your machine? Bring your own Anthropic or Cursor key in the engine chooser, and the conversation and the model traffic never leave the laptop.
Your agent and the platform's own assistant share one craft: the same skills, the same catalog, the same rules about what never happens without your say-so.
What Actually Runs on Your Machine
A full platform as ordinary processes. No containers to babysit.
The Control Plane and One Runner
The same control plane that serves planton.ai, running as a local process, with a single runner that is ready the moment the app is. No slug to name, no registration, no credentials to hand over.
Postgres, Temporal, and a Cache, as Native Processes
Downloaded once on first launch, verified, and supervised by the app. Docker is not required.
OpenTofu and Pulumi as Your Tools
Planton honors the tofu or pulumi already on your PATH. Only when nothing usable exists does it install one, through a version manager you already use or from the tool’s official distribution, checksum-verified.
Your Data Stays Here
Secrets encrypted in the local database with the key in your OS keychain. State on your disk. Logs on your disk. Local data never leaves the laptop.
Cursor for Cloud Infrastructure
Infra Hub on Your Laptop
Describe what you need, watch it compose on a live canvas, see the cloud bill and the IAM policy before anything is created, deploy, and publish it as an Infra Chart — a template your team reuses.
Your Clouds, Found, Not Pasted
Planton reads the cloud logins already on your machine (AWS profiles, gcloud configurations, az subscriptions, kubeconfig contexts) and turns each into a ready connection. Cloudflare and DigitalOcean tokens are detected from your environment and verified against the provider before they are offered. You never paste a key into a form.
700+ Resource Kinds Across 8 Providers
The full catalog, seeded into your local instance at first boot.
A Map of Everything You Run
Accounts, environments, projects, resources, and the services that span them: one living picture, with drill-down to each project’s diagram.
Vercel for Backend, In Your Own Cloud
Service Hub on Your Laptop
Push to GitHub. Your laptop notices, builds the commit in a pod, deploys it to the cluster or cloud you connected, and puts a status on the commit.
35 s
push to run start
30 s
build
34 s
deploy
~800 MiB
build cluster at rest
Measured on an M-series Mac, end to end, unattended: pushed at 07:33:26, the run started at 07:34:01 (waking the stopped build cluster), then a 30 s build and a 34 s deploy. A running, verified deployment inside two minutes of the push.
No Public URL, No GitHub App
A laptop cannot receive a webhook, so the local instance asks GitHub whether the branches it watches have moved, with your own gh sign-in and conditional requests that cost nothing while nothing changed. Each push becomes the same run a webhook would have started.
Builds in a Pod on Your Machine
The build runs on a small build cluster the app sets up inside Docker Desktop the first time you say yes. It sleeps after ten idle minutes and wakes on your next push. Your sign-in enters the build once and is deleted when the run ends.
Deploys Where You Point It
The image goes to the cluster or cloud you connected through the same modules hosted Planton uses, with the same rollout verification; the result lands on the commit as a status your pull request shows.
Building on your own machine needs Docker Desktop. Planton detects it and never installs it. Everything else on this page runs without Docker.
The Same Planton Everywhere
This laptop, planton.ai, or your company's cluster: pick where in the instance switcher. One data model, one set of code paths; only which capabilities run differs. Your laptop can also deploy into your team's Planton, with your cloud credentials still never leaving your machine.
Planton turns your own cloud account into a self-service platform. AI designs the infrastructure, verifies the cost and permissions before anything is created, and publishes it as templates your whole team can deploy. Your services then ship onto that infrastructure straight from Git.
Why It Is Free
Planton Desktop is free for individuals, including commercial use. Not a trial, not a tier with the good parts removed: the full catalog, the wizards, the IaC pipelines, service CI/CD, and secrets management are the core product on every edition.
Here is the business model, stated plainly. Planton makes its money when a team adopts it: hosted seats beyond the free 3, self-hosted licenses beyond the community edition's 5 seats, and prepaid AI credits. The solo developer is not a funnel stage. You are the person this was built to delight, and if you one day bring a team, nothing you built gets redone.
See Full Pricing