Glossary
The words these docs and the apps use, in alphabetical order. Where the apps or the portal use a different word, the entry says so.
The Cloud Viewer agent: a small package, based on Vector, that you install on a server. It sends OS metrics every 30 seconds and journald lines at WARN level or higher, opens no inbound ports and changes nothing. Pro and Team. The apps show its state as Agent connected, Waiting for agent, Agent revoked or Retired; the portal shows Connected, Stale or Offline, and a Retired badge. See Install the agent.
Agent token
Section titled “Agent token”The secret a monitored server uses to send its data and its job reports. The portal shows it once, when you add the server, and stores it hashed. See Install the agent.
Raised on our servers when an alert rule’s condition holds, for example a disk more than 85 % full for 10 minutes. Delivered by push, email digest and alert targets on Pro and Team; a free account has no alerts. See When something needs attention and Alert rules.
Alert target
Section titled “Alert target”A destination that receives every alert when it starts and when it is resolved: a webhook, or on Team also PagerDuty. One target on Pro, up to five on Team, set up by an admin on the portal’s Alerts page. See Send alerts where you work.
Anchored and external targets
Section titled “Anchored and external targets”A check target is anchored when it is attached to something you already monitor, such as a server, a load balancer or a floating IP; its checks then appear on that resource’s detail screen. Any other host is an external target, which the portal offers as Something else on the internet and marks External target. Checks added in the apps are always external, and only the portal can anchor. See Watch a website or port.
A Markdown summary of an alert or a resource: its state, recent history, metrics, costs and links, and never a credential. The iPhone and Android apps share it with Share brief; Copy for coding agent copies it with an instruction block in front. See Hand off to a coding agent.
An uptime check: one test of an endpoint from outside your network, from two independent networks. The apps call the three kinds Web endpoint, Port reachable and Certificate; these docs also say HTTP(S), TCP port and TLS certificate check. See Watch a website or port.
Check target
Section titled “Check target”The host that checks hang on. A free account has 4 targets with at most 8 checks in total between them; the allowance the apps show counts targets. On Pro and Team, a target that is not already a monitored server uses a monitored-server slot. See Watch a website or port.
Claude connector
Section titled “Claude connector”A read-only MCP server at https://api.cloudviewer.app/mcp that lets Claude
query one team’s Cloud Viewer data. Pro and Team. See Ask Claude about your
servers.
Connected project
Section titled “Connected project”A Hetzner project whose read-only token is stored, encrypted, in your team’s account, so our servers sync it every 5 minutes whether the app is open or not. Pro and Team. The apps list connected projects under Team; the portal page is Hetzner Projects. See Connect a Hetzner project.
Dedicated server
Section titled “Dedicated server”A Hetzner Robot server. It is not in the Hetzner Cloud API, so Cloud Viewer sees it only through the agent, and its price is the one entered in the portal. The apps list these under Dedicated servers. Pro and Team. See Add a dedicated server.
Digest
Section titled “Digest”The email digest: one email per team, to one recipient, listing the alerts that started and ended since the previous one. Sent every 4 hours, daily or weekly. Pro and Team. See Send alerts where you work.
Fleet block
Section titled “Fleet block”The ranked rows at the top of the dashboard: up to four rows, then the summary row. The app shows no heading for it. The widget that shows a short version of it is called “Infrastructure health”. See The dashboard.
Fleet enrollment token
Section titled “Fleet enrollment token”A token that lets any number of Hetzner Cloud servers register themselves as monitored servers, for example from cloud-init, Ansible or Kubernetes. It can only register servers and cannot read any data. Created in the portal under Monitored Servers → Add server → Fleet enrollment. Dedicated servers use per-server agent tokens instead. See Install the agent.
Fleet row
Section titled “Fleet row”One line per thing you watch, on the dashboard and in Resources: a status dot, the name, a cost arrow, the month-to-date amount and the strip. Tapping a row on the dashboard expands it in place. See The dashboard.
Free account
Section titled “Free account”A Cloud Viewer account without a paid plan: 4 check targets with at most 8 checks in total, every 5 minutes, 30 days of history, no alerts and no connected project. Its checks pause after 60 days without a visit. See Choose your setup.
Free mode
Section titled “Free mode”Using the iPhone or Android app without an account, with on-device projects only. The apps’ Settings show this mode as Free (client-only). The web app has no free mode. See Choose your setup.
Hand-off notes
Section titled “Hand-off notes”The team-wide instructions that Copy for coding agent puts in front of a brief, for example where your runbook is. An admin sets them on the portal page Agent Hand-off Notes, separately for alert and non-alert shares. See Hand off to a coding agent.
Job report
Section titled “Job report”The result of one of your own scheduled jobs, such as a backup, sent with one
HTTP call or the cloudviewer-report command. It appears in the Jobs
section of the server’s detail screen. Pro and Team. See Get told when a job
fails.
Log context
Section titled “Log context”The warning and error lines a server wrote to journald in the last 72 hours, at most 2,000 per server, shown under its charts. The agent sends them, with obvious secrets masked on the server. The dashboard’s Log activity (24 h) adds them up; the portal page Log Collection controls the optional SSH and sudo lines on Team. Pro and Team. See Log context.
An uptime check status: our two networks disagree about the endpoint. Shown on screen, never alerted. See Watch a website or port.
Monitored server
Section titled “Monitored server”A cloud or dedicated server that runs the agent. Pro has 20 and Team 50, and check targets that are not monitored servers share these slots. The portal page is Monitored Servers. The apps show the agent’s state on the server’s detail screen, and list dedicated servers under Dedicated servers. See Install the agent.
Stops notifications for one alert for 1 hour, 4 hours or 24 hours; the web app offers 1 day instead of 24 hours. A muted alert stays visible and recorded, and is never sent to the digest or to alert targets. Unmute and Extend mute end or lengthen it. See When something needs attention.
No data
Section titled “No data”Grey, on a strip or a check: Cloud Viewer has no result for that time. It means not observed, never down, and it sends no alert. See The dashboard.
On-device project
Section titled “On-device project”A Hetzner project added in the app with a token that stays on the device, in the iPhone’s Keychain or Android’s keystore. The apps call this place On this device on the dashboard, On This Device in iPhone Settings, Projects on this device in Android Settings and This device in the scope picker. See Connect a Hetzner project.
Portal
Section titled “Portal”The website at account.cloudviewer.app where you manage your account, your team, connected projects, monitored servers, checks, alert delivery and the subscription. See Where to use Cloud Viewer.
Push topic
Section titled “Push topic”A team’s private topic that alerts are delivered over. The portal’s Alerts page does not show it yet and explains why. It offers Issue a new topic, for example when someone leaves the team. See Send alerts where you work.
Retired
Section titled “Retired”A monitored server whose cloud server was deleted in Hetzner, or whose agent was removed. It stays visible for 72 hours with a Retired badge, so its logs and job history can still be read, and is then deleted with its data. See Install the agent.
One of the three groups the dashboard ranks rows into: Wrong now (firing, down or degraded), Green, but (fine now, but with trouble in the last 30 days) and Costing more (more than last month, the largest increase first). The apps do not use the word; the three names appear as headings when Resources is grouped by Attention, followed by Everything else. See The dashboard.
Sample data
Section titled “Sample data”A small example project bundled with the apps, added with Explore with sample data on the dashboard or in Settings. The web app shows it whenever you are signed out; its Settings then show the mode as “Demo (sample data)”. See Quick start.
Two meanings. A job is stale when it has not succeeded within the time it
declared (max_age_seconds), which raises an alert. In the portal,
Stale is also a monitored-server badge based on when the server last
reported. See Get told when a job fails.
A row’s history: 30 cells, one per day, oldest on the left, each showing the worst state of that day. Green is reachable, amber degraded, red down and grey no data. The widgets show 7 or 14 cells. See The dashboard.
Summary row
Section titled “Summary row”The last line of the fleet block: how many things you watch, the count per colour and the month-to-date total. It is always shown, and tapping it opens Resources. See The dashboard.
The group that an account’s projects, servers, checks and plan belong to, and also the name of the plan for five seats. People in a team are an Administrator or a Member. Free and Pro have one seat, so only Team can invite anyone. See Bring in your team and Roles and permissions.
Web app
Section titled “Web app”Cloud Viewer in the browser at app.cloudviewer.app. It needs an account; signed out, it shows sample data. See Where to use Cloud Viewer.