# Security: Signed Jobs, Hosting in Germany, SBOM | fliix Ops

> How fliix Ops protects devices and data: signed jobs and downloads, vendor drivers with rollback, read-only AI, German data centres, SBOM.

Canonical HTML: https://www.fliix-ops.com/security

## Hero

Security

**What fliix Ops protects, and how.**

This page describes what the Console and the Windows agent actually do today. fliix Ops itself is not certified to ISO/IEC 27001 or SOC 2; the technical and organisational measures in the data processing agreement are binding.

Security at a glance

- Signed jobs and downloads
- Data centres in Germany
- SBOM for every release
- A way to report vulnerabilities

## Architecture

Architecture

**The agent reports facts, the Console evaluates them.**

The agent collects facts on the device and sends them to the Console over TLS. The rules in the Console decide whether a fact becomes a finding, so new checks need no new program on your devices.

**Every job is signed**

The agent changes something only when the Console sends it a job signed with Ed25519. The job is valid for this one device and until a fixed expiry; without a valid signature the agent discards it.

**The private key stays in the Console**

The agent knows only the public keys. Anyone reading the connection cannot forge a job.

**One credential per device**

Every device signs in with its own credential. An agent sees neither other devices nor other tenants.

## Downloads

Downloads

**Signatures and checksums on every download.**

The agent installs an update only when checksum and signature match. This applies to our own programs and to software from other vendors that the agent installs.

**Signed with a timestamp**

Every program of the agent carries the Authenticode signature of fliix Ops with a timestamp.

**SHA-256 next to every file**

Every published file comes with its SHA-256 checksum, which the agent checks before installing.

**One version, one list of packages**

Every published version stays archived unchanged. If the included libraries change, the upload stops until the version is raised. So the SBOM of a version always matches what your devices receive.

**Third-party software is checked too**

The agent checks the RustDesk client against the stored checksum and the vendor’s signature.

## Drivers

Manufacturer drivers

**Drivers only from the manufacturer, with a way back.**

The agent installs drivers only from the device manufacturer’s catalogue and only from the manufacturer’s servers.

**Checksum and signature before installing**

The agent checks the SHA-256 checksum from the manufacturer’s catalogue and the package signature.

**A way back on failure**

If an installation fails, the agent restores the previous driver.

**No restart without you**

The agent never restarts a device on its own.

**Pilot devices first**

New driver packages go to pilot devices first before they are released for all devices.

## Ai

AI support · Optional · BYOK

**Read-only AI with your own key.**

AI support is optional and runs only with an AI key you store yourself. Without your key, the AI chat about your devices stays off.

**A fixed list of tools**

The AI can call only tools from a fixed list in the Console and run only read-only commands on your devices.

**A checking service without rights**

The text of a PowerShell command is checked by a separate agent service without administrator rights, whose outbound connections a firewall rule blocks.

**Placeholders instead of names**

Before anything goes to your AI provider, the Console replaces names, addresses and identifiers with placeholders.

## Data

Data locations

**Where your data is stored.**

Every sub-processor and its address is listed in the data processing agreement, which you can read in the Console.

**Console, database and files**

Data centres of Hetzner Online GmbH in Germany (Falkenstein and Nuremberg). Hetzner’s German data centres are certified to ISO/IEC 27001.

**Encrypted backups**

Hetzner data centre in Helsinki, within the EU.

**Agent downloads**

Hetzner Object Storage in Falkenstein.

**Support requests by email**

Our own ticket system on an IONOS server in Berlin.

## Access

Encryption and access

**Encrypted in transit, granted by role.**

Every connection to the Console, including the agent’s and the real-time connection, uses TLS.

**BitLocker keys protected twice**

The agent additionally seals recovery keys with a public key of the Console before sending them; they are stored with a separate data key per tenant (AES-256-GCM).

**Roles per tenant**

A person’s role decides what they can see and do. You set up your own roles yourself.

**Sign-in with passkeys**

Passkey or password, optionally with a second factor or only through your Microsoft Entra directory. Before sensitive actions, the person confirms their identity again.

**Four-eyes approval for scripts**

Your own scripts run only in an approved version. With four-eyes approval, nobody can approve their own script.

**Staff administration only via VPN**

fliix Ops staff work in a separate administration that is reachable only through a VPN, and see no BitLocker keys there.

## Sbom

Supply chain

**A software bill of materials for every release.**

For every published version of the Console and every program of the agent we automatically create a software bill of materials (SBOM) in CycloneDX format. It lists every included library with version, licence and origin, so you can check yourself whether a new vulnerability affects fliix Ops.

Index of the agent SBOMs

https://repo.fliix-ops.com/maintenance/windows/sbom/fliixOps.Sbom.json

**In the Console’s Trust Center**

Customers download the SBOM of every version with its SHA-256 checksum in the Console’s Trust Center under “Security”. It is an annex to the technical and organisational measures.

**Agent SBOMs in public**

The SBOMs of the agent programs are public next to the downloads. An index lists every version with its checksum.

**Dependencies kept current**

We check dependencies for new versions every week; security updates are proposed immediately and tested before they are taken over.

## Disclosure

Report a vulnerability

**Found a vulnerability?**

Please write to us if you have found a vulnerability in the Console, in the agent or on our servers.

security@fliix-ops.com

Email

We confirm receipt within 2 working days and get back to you with a first assessment within 7 days.

Please do not access other customers’ data, do not disrupt operations, and give us time to close the gap before you publish it. If you report a vulnerability in good faith under these rules, you need not fear legal action from us.

The address is also published machine-readably in security.txt.

## Faq

Common questions

**Questions about security**

**Q:** Is fliix Ops certified to ISO/IEC 27001 or SOC 2?

No. fliix Ops itself is not certified. The Hetzner data centres in Germany where the Console runs are certified to ISO/IEC 27001. What we do is described in the technical and organisational measures of the data processing agreement.

**Q:** Can someone run commands on my devices through fliix Ops?

The agent only runs jobs the Console has signed; each is valid for that one device and for a limited time. A person’s role in your tenant decides what they can trigger. Your own scripts run only in an approved version, optionally with four-eyes approval.

**Q:** Do I have to use the AI support?

No. AI support is optional and runs only with your own AI key. Without your key, the AI chat about your devices stays off.

**Q:** Where do I get the SBOM?

Customers find them in the Console’s Trust Center under “Security”, for every published version of the Console and every program of the agent, each with its SHA-256 checksum. The SBOMs of the agent programs are also public: https://repo.fliix-ops.com/maintenance/windows/sbom/fliixOps.Sbom.json

## Cta

**More questions about security?**

Write to us. We answer questions about architecture, data protection and data processing directly.
