Deploy With One Command

Your laptop, your AWS account or your Kubernetes cluster. Same workflow, same CLI.

One command starts your agents and gives you an endpoint and a dashboard.

Deploy locallyRead the docs

The Problem It Solves

Moving a workflow from your laptop to the cloud shouldn't mean rewriting it

A multi-agent app that works on a laptop usually needs a second life to run anywhere else: Dockerfiles per agent, a cluster, manifests, scaling rules, and a different setup for each cloud. Every move means new files to keep in step with the code. With CanyonOS the workflow stays the same; one setting decides where it runs, and one command does the rest.

One Workflow, Three Places to Run It

1. Pick where it runs.

Your laptop is the default. To run on AWS or Kubernetes, change one setting.

2. Run one command.

CanyonOS checks your project, packages each agent and starts your workflow.

3. Use it.

You get an endpoint to call and a dashboard to watch your agents. One command takes it all down again.

$ canyonos deployyour laptopyour AWS accountyour Kubernetes cluster

Deploy Locally or in the Cloud

Deploy locallyRuns on your laptop with Docker. Nothing to set up. Local guide
Deploy to the cloud or KubernetesRuns in your own AWS account or on your Kubernetes cluster. Change one setting. AWS guide · Kubernetes guide

For the People Who Run It

Where the agents runYour laptop, your own AWS account, or a Kubernetes cluster you already run.
What it creates in AWSA machine to deploy from, an access role with only the permissions it needs, and network rules. You see everything before it's created, and one step removes it all again.
Side by sideSeveral workflows can run on one machine without getting in each other's way.
What you addDeployed endpoints have no built-in authentication, so put your own in front.

What 0.2 Doesn't Do Yet

CanyonOS 0.2 is our first public release, and we ship fixes every week. Here's what it doesn't do today:

  • Outside Kubernetes, AWS is the only cloud it launches machines on.
  • On AWS, only x86 machines are supported, and outside us-east-1 you supply your own machine image.
  • Deployed endpoints have no built-in authentication.