Working with SnapVM
SnapVM is built around a small set of durable concepts: machines, terminal sessions, services, storage, and organizations. The web console is the primary way to operate them.
Typical workflow
Section titled “Typical workflow”- Choose an organization in the console.
- Create a machine for a project, demo, class, or agent host.
- Open a terminal and set up the workspace.
- Start short-lived exploratory commands in the terminal.
- Define services for long-running processes.
- Stop or leave the machine to hibernate when idle.
- Reopen the machine later and continue from the persisted filesystem.
Machines
Section titled “Machines”A machine is a named workspace backed by a runtime. It owns the state you care about: files, packages, repositories, service definitions, and gateway routing.
Machine names also become part of the application URL. Keep names short and stable, for example api-demo, agent-lab, or frontend-review.
Sessions
Section titled “Sessions”Terminal sessions are interactive shells exposed by the machine agent. They are useful for development, debugging, installing tools, and inspecting logs.
Sessions can survive browser disconnects, but they are still interactive work surfaces. If a process must keep running across hibernation, make it a service instead of relying on an open terminal tab.
Services
Section titled “Services”Services are durable process definitions. Use services for web servers, databases, queues, workers, language servers, or agent processes that should restart when a machine wakes.
A service usually has:
- A name
- A command and arguments
- A working directory
- Optional environment variables
- An optional HTTP port
- Lifecycle state such as running, stopped, or failed
Storage and hibernation
Section titled “Storage and hibernation”The filesystem persists across stop and hibernation. This includes source code, installed packages, databases stored on disk, configuration files, and generated artifacts.
In-memory state does not persist. Processes started in an interactive shell stop when the runtime sleeps unless they are managed as services.
Organizations and access
Section titled “Organizations and access”Machines belong to organizations. Authentication and row-level security keep organization data isolated so a user only sees machines for organizations where they have access.
Common access tasks happen in the console:
- Create or switch organizations
- Invite members
- Review machine ownership
- Update member roles when the deployment supports role management
Choosing the right primitive
Section titled “Choosing the right primitive”| Need | Use |
|---|---|
| A durable workspace for a project | Machine |
| A browser shell for exploration | Terminal session |
| A long-running process | Service |
| Files and packages that survive idle time | Persistent filesystem |
| Team-level ownership and access | Organization |