Privacy policy

This page explains which personal data Nixorium processes, why, and for how long. Nixorium is an open-source project. Its online services are operated by HardyPress SRL (VAT number IT06749370489), the data controller for the processing described in sections 1–4. Contact: [email protected]. Full company details are in the terms of sale.

Nixorium itself runs on your school’s computers. The software you install does not send us anything unless you enable one of the optional services below.

1. Website

• Hosting. The site is served by Cloudflare. Cloudflare processes connection data, including IP addresses, to deliver pages and protect the service.
• Statistics. We use Cloudflare Web Analytics, which does not use cookies and does not track you across sites. We see aggregate page views only.
• Installer counter. Requests to nixorium.org/install.sh are counted per day. We store only the date and the count, never your IP address.

2. Controller reports (optional)

If an administrator enables reports on the controller, it sends a small monthly-rotating report: Nixorium version, deployment mode, configured lab size and similar technical counters. The report contains no names, no addresses and no content from the laboratory. Individual reports are kept for about 90 days; aggregate monthly counts, which contain no identifiers, are kept for 24 months. Reporting can be disabled at any time in the administrator interface.

3. Nixorium Assistant (optional, subscription)

Assistant is an optional service that helps a laboratory administrator change and troubleshoot the laboratory configuration. It only works after the administrator activates it with a code and accepts a consent screen. Versions that support diagnostic logs and configuration history require renewed acceptance of the updated screen before sending this additional data.

What is sent when the administrator uses Assistant

• the messages written in the Assistant chat;
• the contents of the laboratory configuration files that Assistant reads while preparing a proposal, and the results of its checks (for example evaluation errors);
• technical context: Nixorium version and the list of configuration files;
• when Assistant checks the laboratory state, the same information shown in the administrator overview: computer names and laboratory addresses, whether each computer is reachable, the installed and expected configuration revision, pending updates and the result of the last update;
• on versions that support diagnostics, fixed read-only observations from declared laboratory computers: the current configuration revision, service state, free space, reset and update indicators, and bounded excerpts from current-boot logs for a predefined list of Nixorium and system services;
• on versions that support configuration history, bounded saved Git commit metadata and visible configuration file changes within the ancestry of the saved configuration used for the conversation.

Before configuration contents and tool results leave the controller, filters remove known secret patterns, such as recognized password hashes and private keys, and credential fields they recognize. Diagnostic excerpts also filter common private identifiers. Filtering is best effort: personal or sensitive text may remain in configuration contents, commit messages or log excerpts. Diagnostic tools do not read student home files.

Please do not write personal data about students or staff in the chat. The laboratory configuration normally contains no personal data about students.

Who processes it

• Our service (on Cloudflare) forwards each request to the AI model and returns the answer. Configuration files and tool results are not kept as separate records unless the administrator reports or shares a conversation as described below. It keeps the text of messages and answers for 30 days to detect abuse, then deletes it. These texts may quote configuration, history or diagnostic observations. It also stores technical metadata about each request: time, number of steps, token counts, cost and outcome. We keep this metadata for 24 months, then delete it.
• The AI model is accessed through OpenRouter and run by the model provider (currently OpenAI, model GPT-6.1 Sol; OpenRouter, Inc. is based in the United States). We configure OpenRouter to route requests only to providers that do not use the data for training. The model provider (currently OpenAI) may keep requests for up to 30 days to detect abuse, then deletes them. These providers may process data outside the European Union, on the basis of the European Commission’s Standard Contractual Clauses and, where the provider is certified, the EU-U.S. Data Privacy Framework.

Feedback and conversation sharing (optional)

The administrator can choose to send us a conversation, to help us improve Assistant:

• Report this answer: after an answer, the administrator can mark it as helpful or wrong and send that conversation, after a confirmation screen, each time.
• Share conversations: an optional setting, off by default, that sends every conversation of that laboratory.

In both cases we receive the conversation messages and tool results, including the configuration contents Assistant read, its check results and any configuration history or diagnostic observations it received. The controller’s filters described above apply; personal or sensitive text may remain. We use these conversations only to find and fix mistakes in Assistant. They are stored privately, accessible only to the operator, and deleted after 90 days. The legal basis is the consent given by the administrator, which can be withdrawn at any time by turning sharing off; reports already sent can be deleted on request.

Account data

To create a trial or a subscription we collect the school name and an email address. We store a hash of the activation code and of the laboratory’s access token, never the code or the token itself. We use the email address only to send the activation code and service messages about your subscription.

The registration form uses Cloudflare Turnstile to block automated abuse. Turnstile checks the browser without tracking cookies; Cloudflare processes the connection data needed for the check.

Free trial. We keep the email address of a trial for as long as the account exists, to tell laboratories apart and to limit how many codes an address can receive.

4. Payments and invoices

• Stripe processes payments and calculates taxes. We do not see or store card numbers.
• Fatture in Cloud (TeamSystem) prepares electronic invoices and sends them to the Italian Exchange System (SDI).
• Postmark sends our transactional emails.

Billing and invoice data are kept for the period required by Italian tax law (currently ten years).

5. Legal bases

• Website and installer counter: our legitimate interest in running and protecting the site.
• Controller reports, Assistant feedback and conversation sharing: your choice to enable or send them (consent), withdrawable at any time.
• Assistant, account and payments: performance of the contract you request.
• Keeping the text of Assistant messages and answers for 30 days to detect abuse: our legitimate interest in the security of the service.
• Invoices: compliance with legal obligations.

When a school uses Assistant, the school is the data controller for any personal data contained in what its administrator sends; we act as its data processor under a separate data processing agreement, available on request.

6. Your rights

You can ask for access, correction, deletion, restriction or portability of your data, and object to processing, by writing to [email protected]. You can also lodge a complaint with the Italian Data Protection Authority (Garante per la protezione dei dati personali, www.garanteprivacy.it). On request we delete an account’s personal data; invoices are kept for the period required by tax law.

7. Changes

We will publish changes on this page and update the date above.