Server map: what RootPilot sees when you connect

Until now RootPilot knew little about a server on connect: the OS, resources, a few packages. Now a single script takes one pass and collects an inventory: server software and versions, services and their state, open ports and what listens on them, containers, timers and cron, certificates, users with sudo, SSH and firewall settings.

The inventory becomes a map: the server’s role — web, database, reverse proxy, application — what proxies to what, what faces the internet and what is local only. The map is shown in the “Infrastructure memory” panel, and the agent reads the same map before every task: it does not have to rediscover which web server runs here or where its config lives. A rescan records only changes to the log — a new port, a different version, a service that disappeared.

In an empty chat the server leaves a note. Each item is a fact from the map and what it means: “SSH on a public port accepts password login”, “PostgreSQL present, no backup job scheduled”, “certificate expires in 9 days”. Next to it — what is fine. Any item can be handed to the agent with one button, or marked as not relevant for this server: password login may be your deliberate choice, and the note will not bring it up again.

The note collapses to a single line and never expands on its own. This is an inventory check — no commands are run, nothing watches in the background. The map says what is on the server. Checking whether it actually works — say, whether the backup is fresh — is still a task you hand to the agent.