How security works

The agent is installed only with your approval, every command is signed, and the person at the computer sees the session. Below are all the measures and their limits.

How the agent connects

The agent connects out to our server by itself over TLS: no ports are opened on the computer, and no VPN is needed. The agent and the installer accept a certificate only if it is issued for our server's name by one of the certificate authorities on a list built into the program: ISRG (the root authorities of Let's Encrypt) and Google Trust Services. A certificate from anyone else, for example from a proxy or an antivirus that decrypts HTTPS, is rejected, and the agent does not connect.

Installed only with your approval

The installer warns that it is installing remote management software and won't let you skip the warning for 10 seconds: do not install if a stranger or “tech support” asked you to. After that, it needs your install key.

In the panel, an “Agent install request” window appears with the key and the IP address the request came from. The “Allow install” button becomes active after 10 seconds; “Deny” turns the request down. Only after approval does the agent receive its credentials, once, over a protected channel; they cannot be collected a second time.

The person at the computer sees the session

While the agent is running, a session window is open at the edge of the screen. It shows which modules the administrator has used (for example, the terminal or files); the commands themselves are not shown in the window. The window cannot be closed: if it is terminated, the agent opens it again a few seconds later. There is no hidden mode.

In the session window, the person can pause the session and resume it (a pause lasts until the agent restarts) or send a report. In the tray icon menu, they can switch remote access off completely until they turn it back on themselves, or uninstall the agent. While the session is paused or access is off, the agent does not carry out a single command.

When a technician works on the computer, the window shows their name in front of the module; once your account has technicians, your name is shown too.

Sending a report

From the session window, the person can send a report with a reason (administrator actions, technical issue, unauthorized access or other) and, if they wish, a description. The report comes to us together with the record of actions on that computer, and we look into it. The account owner sees reports in “Devices”, except reports of unauthorized access: only we see those. If it is clear that the computer was managed without the right to do so, we may disconnect the computer and suspend the account. Reports are shown to the account owner, not to technicians, and the report we receive says who worked on the computer in the last 15 minutes.

Every command is signed

Each computer has its own secret key. Every command, including a stop, is signed with that computer's key and contains the time and a one-time value: the agent rejects a command that is altered, unsigned, outdated or replayed. A key taken from one computer is of no use for the others. The agent's credentials are tied to the fingerprint of the computer they were first used on, so simply copying them to another machine won't work.

Only signed code runs

The code of the modules that run on the computer is signed with an Ed25519 key. We do not sign it on the server: only the public key is kept there. The agent checks the signature against the key built into it and does not run unsigned or altered code. So even someone who takes over the server cannot swap the code of the modules themselves. What they can do through the existing modules is described below, in “If our server is hacked”.

Protection on the computer itself

The agent is installed in Program Files, where only administrators can write. The file with its credentials is closed to everyone except the system and administrators, and is encrypted with the built-in Windows protection (DPAPI) for the user account the agent was installed under.

Account protection

Each password sign-in is confirmed with a six-digit code sent by email; you can also sign in with Google. Your profile shows open sessions and the sign-in history: you can end any session, or close all the others with one button. The panel opens only with a paid plan or set.

Technicians in your account

You can invite technicians by e-mail to work in your account, each with their own login. Only you invite and remove technicians and change their rights. Rights are closed by default: you choose which groups of computers and which modules each technician may use, and the server checks every command against them. A technician opens your panel only after adding two-step sign-in (an authenticator app or a passkey) and only from a session that passed it. Answers from a computer reach only the person who sent the command, and you.

Rights are confirmed again every minute: when you remove a technician or change their rights, it takes effect within a minute. Everything a technician does goes into the action log under their name, and you see who did what in the Activity tab. The person at the computer sees the technician's name in the session window.

If our server is hacked

To put it plainly: whoever takes over the server the commands pass through will be able to send module commands to the computers that are online, including to the terminal, and the terminal runs any command. There is no full protection against this in such a setup, and we do not hide it.

What still holds: the code of the modules themselves cannot be swapped (the signing key is not on the server); the person at the computer can pause the session or switch access off, and then commands are not carried out; log records made before the takeover will remain in the copy outside the server, although whoever took over the server may not log the new actions.

What we store

For each computer we store its name as the agent reports it (user@computer), the IP address and time of its last connection, and the IP address of each install request. For the statistics we also store each computer's load over time (processor, graphics card, memory, disks and network) and, for every hour, the names of the few programs that used the most resources; this is kept for up to 30 days and then deleted automatically. Once a day the agent also reports which antivirus and firewall are on and up to date, the health of each disk, the Windows updates installed in the last 62 days (their numbers and dates), the Windows version and the memory size; we keep the latest such report. For monthly reports we also keep a short summary of each day for every computer: time online and in use, average and peak load, how full drive C is, that day's alerts, the program that used the processor most, the state of the antivirus, firewall and disks, the updates installed, the Windows version and the memory size, and on plans with Windows policies also how many updates are waiting. These summaries, and any note you write into a report, are kept for up to 13 months. If you send reports to your clients, we also keep the addresses they go to (until you remove them or a recipient stops the e-mails), the name, contact line and logo you put on your reports, the links to reports (60 days, counting only how often each was opened) and a log of when a report went out and to how many addresses (12 months). On plans with Windows policies we also store each computer's Windows edition and build, how many updates are waiting and whether it needs a restart; if you set policies, also the settings you choose, how each one stands on every computer, and a record of policy changes, kept for 12 months. Agent tokens are stored only as hashes, and each computer's keys are stored encrypted. Screenshots, files and command output pass through Cloudflare and our server to your browser and are not stored anywhere; in transit they are protected by TLS, but there is no end-to-end encryption from the computer to the browser. The servers are in Oracle Cloud in Frankfurt (EU), and traffic goes through Cloudflare. Payments are processed by Freemius, which never sees any data from your computers. If you invite technicians, we also store the e-mail addresses you invited, each technician's name, rights and the dates they joined and left, and a record of team events with the e-mail addresses and IP address, kept for 12 months.

Action log

We keep a log of administrator actions: which command was sent to which computer and when, install approvals and denials, changes to install keys and devices. The commands themselves are recorded too, for example the text of a terminal command if it is no longer than 1024 characters, so do not put passwords into commands. The contents of screenshots and files, as well as command output, do not go into the log; for uploaded files, only the size is recorded. Each record is linked to the previous one through a hash, so the log cannot be altered unnoticed, and a copy of it is kept outside the main server, in storage where records can only be added. The log and reports are kept for 12 months. Each record shows who acted: you or a technician by name.

Deleting data

When an account is deleted, its devices, agent credentials, install keys and install requests are deleted immediately. Your technicians lose access at once, and their rights and your groups of computers are deleted with the account. The action log and reports, as security records, are kept for 12 months from the moment they are created and are then deleted automatically. All other data is deleted within 14 days of a deletion request or the end of service, and disappears from backups within 30 days. When a technician deletes their own account, only their place in teams ends; the records of their actions stay in the owners' logs for the same 12 months.

Found a vulnerability?

Write to [email protected] with what you found and how to reproduce it. We will reply.

AxXtori