How Valinor works
From an account to the first resolved incident, in seven steps. Each animation follows the screens of the dashboard; the data in them is example data.
You create an account
Start free with one external site, or try Starter or Plus for 14 days. No card.
- Pick where you start: Free watches one site from the outside. Starter and Plus start with a 14-day trial and get the agent on the server.
- Create the account: With email and password, or with Google. No card is asked for, on Free or during the trial.
- Open the dashboard: The dashboard is empty until you add the first host. Everything starts from here.
You add the first host
Four host types, four ways to connect them. Pick yours.
VPS with agent
- Name it and pick the type: In the dashboard, under “Add a host”. Pick “VPS with agent” and how often the agent reports.
- Copy the command: The command carries the host’s key. The secret is shown once, so that is when you copy it.
- Run it on the server, as root: The installer downloads the agent, checks its fingerprint, starts a timer and sends the first report.
- The host shows up: The first report is accepted during the install. The outside checks are scheduled too.
Shared hosting
- Pick “Shared hosting”: A name and the account’s domain. The agent finds the webroot by itself.
- Copy the same command: It is the same as on a VPS, with this account’s key. The secret is shown once.
- Run it from the account terminal: From cPanel → Terminal or over SSH, as the account’s user. No root: the agent adds itself to the crontab.
- The account shows up: Fingerprints of the webroot files, malware patterns and the outside checks on the domain.
External site
- Type the domain: Optionally, a text the page must contain. Nothing gets installed.
- The HTTP probe answers first: Within the first minute you see whether the site answers, and with which code.
- The other checks get scheduled: TLS, DNS, blocklists, headers, email, domain expiry: each at its own pace.
cPanel/WHM reseller
- Give the WHM address and an API token: You generate the token in WHM, under Manage API Tokens. Nothing gets installed on the server.
- Save and test: The dashboard asks WHM which accounts exist and tells you at once whether the token works.
- One host per account: The list is read again every 30 minutes, so new accounts show up by themselves.
The checks run by themselves, each at its own pace
Not everything that matters changes equally often. So there are three paces, not one.
- The agent’s report: Every 5 minutes by default; you pick between 1 and 60. It brings the logins, processes, firewall, disk and site files.
- The HTTP probe: From the outside, at your plan’s interval. It is the first to say a site stopped answering.
- The slow checks: The TLS certificate, DNS, blocklists, domain expiry: from 15 minutes to 12 hours.
Related signals become one incident
When something fails, several things fail at once. You get one alert, not one for each symptom.
- The signals arrive: The HTTP probe gets a 502 on two sites, and the agent’s report finds the web server stopped.
- They gather into one incident: Signals that belong to the same problem go into the same incident, with their timeline.
- One alert goes out: By email and push; on paid plans, in Slack or Discord too. What follows is added to the incident, with no new alert.
- Findings stay on the host page: A missing security header or an incomplete SPF does not wake you at night. You find them on the host’s page.
You ask for the AI analysis when you want it
The analysis does not start by itself. You ask for it from the incident, and it uses one AI message from your plan.
- Open the incident: The timeline and the evidence are already there: what was seen, when and from where.
- Ask for the analysis: One click, one AI message. Paid plans include 30 a month, and Free 10.
- Get the probable cause: In plain words: what happened and what to do. If a fitting remediation exists, it proposes it and waits for your approval.
A remediation runs only after you approve it
By default the agent only reads. For a command to run, two switches have to be on, and then each remediation is approved one by one.
- The switch on the server: It is turned on only at install, with --enable-actions. An agent update cannot turn it on.
- The switch in the dashboard: On the host’s page. Without both, the agent refuses any command, however correctly it is signed.
- The remediation waits for you: The animation stops here: the next step is yours. Approve or reject, right in the window.
- The agent picks it up: The signed command leaves with the agent’s next report. The configuration is tested before the restart.
- The result goes into the incident: The HTTP probe confirms the site answers, and the steps stay in the incident’s history.
What stays after an incident
The incident closes, but its trace stays. A month later you can still answer “what happened back then?”.
- The problem log: Every problem of the host, with a start, an end and an explanation in plain words.
- The uptime history: Every outage keeps its reason, for as long as your plan’s history goes.
- The reports: A summary for a period, which you print or export as CSV.
Ready for your first host?
An external site is free to add, no card. The agent on a server comes with Starter or Plus, with a 14-day trial.