Skip to Content

Endpoint agent

Install, update and remove the endpoint agent on Windows, macOS and Linux, and what it reports from the device.

RequiresDevice Identity (DIAM)

Add-on required: The endpoint agent is part of the DIAM (Device Identity) add-on.

The endpoint agent is the software you install on each endpoint. It enrolls the machine, reports its security posture on a schedule, and carries out the remote actions you trigger from the console. Without it a device cannot appear in the inventory. Devices enrolled with the previous Windows client keep working; see Legacy client for what differs and how to move them over.

Supported platforms

PlatformSupported from
WindowsWindows 11 and Windows Server 2019, x64
macOSmacOS 12 Monterey, Intel and Apple silicon
LinuxDebian 11, Ubuntu 22.04 and Red Hat Enterprise Linux 9 (including Rocky and Alma), x64 and ARM64, with systemd

Older versions are not supported. Where the machine has a TPM 2.0 (Windows, Linux) or a Secure Enclave (Apple silicon, Macs with a T2 chip), the agent keeps its device key in it; elsewhere it uses a key stored on disk.

What gets installed

PlatformPackageWhat runs
Windows.msiThe service DTS Endpoint Agent (DTSAgent), installed under %ProgramFiles%\DTS\Agent, started at boot. The agent window starts at every user sign-in.
macOS.pkgA system daemon under /Library/DTS/Agent and the app DTS Endpoint Agent in the menu bar of every user session.
Linux.deb, .rpmThe service dts-agent.service and a status icon in desktop sessions. Servers without a desktop run the service alone.

The agent keeps its state, including its device identity, in %ProgramData%\DTS\Agent (Windows), /Library/Application Support/DTS/Agent (macOS) or /var/lib/dts-agent (Linux), so an in-place upgrade does not re-enroll the machine. Its log files are in the logs folder there, kept for seven days.

Requirements

  • Administrator or root rights to install, because the package registers a system service.
  • Outbound HTTPS to your platform instance. The agent dials out and keeps the connection open. No inbound firewall rule is needed.
  • A proxy that inspects TLS works once its root certificate is trusted in the operating system's own store (the machine store on Windows, the System keychain on macOS, the distribution's CA bundle on Linux). Nothing is configured on the agent.

Get the installer

The installer is built for your instance: the platform address and the certificate the agent needs to enroll are embedded in it, and it is code-signed. An installer from a different instance will not work, and there is nothing to configure at install time.

  1. Open Device Settings → Registration.
  2. Request a build if none exists yet, then download it once it is ready. The card shows the version and its SHA-1 and SHA-256 checksums.
  3. Verify the checksums and distribute the package to your endpoints.

The console currently offers the Windows package. The packages for macOS and Linux are not yet available for download in the console.

See Device settings for the enrollment certificate that anchors this and for rotating it.

Install and enroll

Install the package with administrator rights, by hand or through your software distribution or MDM. It installs for all users and starts the service. The packages take no options, so a silent installation needs nothing beyond the platform's usual switches:

msiexec /i dts-agent-<version>.msi /qn
sudo installer -pkg dts-agent-<version>.pkg -target /
sudo apt install ./dts-agent_<version>_amd64.deb
sudo dnf install ./dts-agent-<version>.x86_64.rpm

Enrollment then happens by itself. The agent creates its device key, registers with the platform and receives its device identity. Nothing has to be entered on the endpoint. Depending on the Device enrollment mode in the device settings, the device is managed at once or waits for your approval in the device inventory.

On macOS, some of the posture (for example parts of the security settings) needs Full Disk Access, which a package cannot grant. Grant it to the agent with a privacy preferences (PPPC) profile from your MDM; without it those values are missing and the rest is reported as usual.

Note: The device identity is created at first enrollment and stored on the machine. Re-imaging a device, or wiping that state, produces a new device record. The old one stays in the inventory until it ages out or you delete it.

What the agent reports

Posture is collected every five minutes and sent as a difference against the last report, with a complete snapshot at least once a day or whenever the platform asks for one. Between reports the agent sends a short check-in every few seconds, which is what makes a device show as connected.

The agent reports:

  • Operating system and hardware: version, build, architecture, computer model, CPU and memory, whether the machine is virtual.
  • Identity of the machine and its users: hostname, domain, the account currently signed in, and the user accounts of the device.
  • Network configuration: adapters with MAC and IP addresses, gateways and DNS servers.
  • Volumes and disk encryption (BitLocker, FileVault, encrypted Linux volumes).
  • Installed software with vendor, version and install date.
  • Security state: firewall, Microsoft Defender on Windows, and settings such as Secure Boot, TPM, Gatekeeper, System Integrity Protection, whether the Mac is managed by an MDM, and whether a password is required after the screen locks.
  • Missing operating-system updates with severity and whether a reboot is required.
  • Whether the console session is locked, with the check-in. The device page shows it only when you enable it.
  • Its own state: version, isolation status and MFA enrollment state.

It collects no geolocation on the endpoint (the platform derives approximate location from the connection's IP address), and no browsing history, file contents, keystrokes or screenshots. The process list, with command lines and the account each process runs as, is only collected when an administrator runs that remote action.

The collection scope is fixed. It is the same on every device of a platform, and users can read it in the agent window (below). Plan the rollout with whoever is responsible for works-council agreements or data-protection sign-off in your organization: the inventory, the signed-in account and the on-demand process list are personal data in most jurisdictions.

The agent window

Every user sees the agent in the notification area (Windows), the menu bar (macOS) or as a status icon (Linux). Its menu shows the time of the last report, and Check in now asks the platform for a fresh compliance result. The window has these pages:

PageShows
StatusWhether the device is connected, enrolled and reporting, and anything that currently fails.
ComplianceThe device's last compliance result with the checks that failed.
Two-factor sign-inOn Windows, the user's second factors for the Windows sign-in, when the device verifies them itself.
PrivacyEvery query the agent runs and the data it last sent.
AboutThe version, and Send diagnostics / Save diagnostics for support cases.

The command line

The agent installs a command dts-agent for administrators and support on the device:

CommandPurpose
dts-agent statusThe agent's state as JSON: connection, enrollment, the last report, isolation, and every step that currently fails with its message. The first thing to look at when a device misbehaves.
dts-agent compliance --refreshThe last compliance result, asked again from the platform.
dts-agent revert-isolationReleases the network isolation locally, run as administrator or root. For a device the isolation has cut off from the platform. The release is logged.

Behaviour when the platform is unreachable

The agent keeps retrying with growing intervals of up to five minutes and reconnects on its own. Posture reports are not queued while disconnected, so after an outage the platform sees the current state, not the missed history. The device appears offline in the console about ten seconds after its last check-in. A full-tunnel VPN that routes everything through the corporate network must allow the connection to your instance too, otherwise the device stays offline while the VPN is up.

Update

The agent does not update itself. An update is a new package, deployed the same way as the first one.

  1. Request the new version under Device Settings → Registration and download it.
  2. Deploy it to the endpoints with your software distribution or MDM.

The installation upgrades in place and preserves the device identity, so devices do not re-enroll and their history is kept. Test the upgrade on a pilot group before a fleet-wide rollout.

Uninstall

  1. Uninstall the agent through the platform's usual way (Settings → Apps on Windows, your software distribution or MDM, apt remove or dnf remove on Linux). On macOS run sudo dts-agent uninstall. The uninstall releases a network isolation first, then removes the service, the program files, the agent's state and its device key.
  2. Delete the device in the inventory. Uninstalling does not notify the platform, so the record stays until you remove it.

Note: On Debian and Ubuntu, apt remove keeps the agent's state until you purge the package (apt purge). Removing the .rpm, and the uninstall on Windows and macOS, remove it.

Troubleshooting

  • The device never appears in the inventory. Run dts-agent status on the device: it names the step that fails. Check that the endpoint can reach your instance outbound, and that the enrollment mode accepts new devices. An installer built for a different instance never enrolls.
  • The status reports a connection error, but the network works. Usually a proxy that inspects TLS, whose root certificate is not in the operating system's store. See Requirements.
  • A device appears twice. The machine was re-imaged or lost its local state and enrolled again. Delete the old record.
  • Software or patch data looks stale. Posture is reported on a schedule. Use the Request full update remote action to force a complete report.
  • The browser asks for permission to access the local network during sign-in. Compliance checks at sign-in work by having the login page ask the agent on the machine itself. Chrome 142 and newer, and Edge, prompt the user before a website may reach a local address. Users can allow it once per site. On managed fleets, allow it centrally for your sign-in address with the browser policy LoopbackNetworkAllowedForUrls, otherwise every user sees the prompt.
  • A device the isolation cut off can no longer be released from the console. Run dts-agent revert-isolation on the device as administrator or root.

See also

On this page