Diagnosis

A Dusk node lets you see and control the work Dusk itself is running on a device, live and over the wire - the same way docker ps / docker logs / docker exec show and drive containers rather than the host. This is the Dusk layer you embedded, not the host operating system's process table.

What you can do today

• See Dusk's processes - ps lists the processes running inside the node (the Dusk programs that have been started), with each one's pid, name, and version. Like docker ps, it shows Dusk's own process table, not the host's.
• Read logs - logs opens the node's log buffer in an interactive viewer, or streams it to a file, otel collector, or just an http server. See Logs.
• Run commands - open the shell and run Dusk programs against the node interactively.
• Stop or restart work - kill signals a Dusk process by pid; with the shell that's enough to recover a stuck workload.
• Drive it programmatically - the Python API and the API gateway do all of the above from a script, over HTTP, or from an AI agent.

What you can see and do is exactly what the node's programs expose. The Base programs cover the Dusk layer itself (see that page for each program's full usage); to surface something host-specific (a sensor, a hardware register, an app-internal metric), that's a program you write.

The interfaces

Diagnosis happens through whatever client is handy: the interactive shell for a person, the Python API for a script, or the API gateway for anything over HTTP, an agent included. They all drive the same node.