For AI agents: The documentation index is at https://docs.digitalocean.com/llms.txt. Markdown versions of pages use the same URL with index.html.md in place of the HTML page (for example, append index.html.md to the directory path instead of opening the HTML document).

DigitalOcean MicroVMs are lightweight virtual machines built for workloads that alternate between bursts of work and long idle periods, such as AI agent sessions, sandboxes for untrusted or generated code, and per-user environments. Each MicroVM runs one container image in its own virtual machine with its own kernel, isolated from other workloads.

You choose the image, size, networking, ports, environment variables, and tags when you create a MicroVM. After that, the platform pauses and resumes it with your traffic, and you can checkpoint it, run commands in it, or destroy it. You manage MicroVMs from the Control Panel, doctl, or the API.

Container Images

A MicroVM runs a public container image, such as one from Docker Hub, or a private image from your team’s DigitalOcean Container Registry (DOCR). Images must be built for linux/amd64.

Auto-Pause and Auto-Resume

A MicroVM pauses after an idle timeout, which defaults to 5 minutes and can be up to 4 hours. The next HTTP request, command, or console session resumes it, with its memory, files, and running processes intact. You can turn auto-resume off at creation to require a manual resume, and you can pause and resume a MicroVM yourself. A paused MicroVM incurs storage charges only. For details, see MicroVM Lifecycle.

HTTPS Endpoints

Each MicroVM gets its own HTTPS URL, authenticated with a DigitalOcean API token. It serves HTTP or HTTP/2 on the HTTP port you choose at creation. For details, see How to Send Requests to a MicroVM.

Public and VPC Networking

A MicroVM uses public networking with a shared NAT gateway by default, or you can place it in a VPC network in the same region. For details, see MicroVM Networking and Access.

Checkpoints

A checkpoint captures a MicroVM’s memory and disk. You can create new MicroVMs that start from that state, already warmed up. For details, see MicroVM Checkpoints.

Commands and Console

You can run a single command or open an interactive shell in a MicroVM’s container with doctl or the API. For details, see How to Run Commands and Open a Console in a MicroVM.