Privacy Policy

Last updated: September 8, 2026

This page explains what data flows through LumaBrowser and lumabyte.com, where it goes, and what you can do about it. The app itself is deliberately quiet: it has no third-party analytics or crash reporting, no account, and no paid plan. The only things LumaBrowser sends to us are a lightweight version check-in and, if you use the Template Builder, the sanitized page templates it generates. This one policy covers the app and the website.

On this page

  1. Who we are
  2. Summary at a glance
  3. What we collect
  4. Page templates, in detail
  5. How we use your data
  6. Third parties we rely on
  7. Data retention
  8. Your rights
  9. Children's data
  10. International transfers
  11. Changes to this policy
  12. Contact

1. Who we are

Lumabyte, LLC is the data controller for information processed in connection with LumaBrowser and lumabyte.com. We're a US limited liability company. When this policy says "we," "us," or "Lumabyte," it means Lumabyte, LLC.

You can reach us at the address in Section 12.

2. Summary at a glance

LumaBrowser runs on your machine. Local models process your prompts and responses on your computer. If you connect cloud models, browse websites, or use external tools and webhooks, the data needed for those requests goes to the services you use. Version check-ins and Template Builder sharing are described below.

No third-party analytics or crash reporting, anywhere. We removed PostHog and Sentry from LumaBrowser entirely, and the lumabyte.com website carries no analytics or crash-reporting scripts either. What does leave your machine is narrow and documented: a periodic version check-in to lumabyte.com (so we can count installs and see version and platform spread), and sanitized page templates if you use the Template Builder.

No account, no payment, no license. LumaBrowser is free with no paid plan, so we hold no billing data, no license keys, and no machine fingerprints tied to a person. There is nothing to sign up for and nothing to cancel.

3. What we collect

We group the data we touch into buckets. This is the full picture across both the app and the site.

BucketWhat it containsWho sees it
App version check-in
(on in packaged builds; off in developer mode)
A small periodic ping to lumabyte.com/api/pulse (roughly every five minutes while the app is open) so we can count how many installs exist and are currently active, and see version and platform spread for release planning. It carries only: the app version and the platform (e.g. win32, darwin, linux) in the body; a stable machine identifier in a header so the same install is not counted twice (the OS-level machine id: on Windows the MachineGuid registry value, on macOS the IOPlatformUUID, on Linux /var/lib/dbus/machine-id); a short-lived rolling authentication token; and a User-Agent string with the version, architecture, OS release, and Electron version. Nothing else. No events, no feature usage, no URLs, no page content, no prompts, no credentials. The machine identifier is not linked to a name, email, or account, because there are none; it is an anonymous counter key. Lumabyte
Page templates
(only if you use the Template Builder)
The target domain and URL path (no query strings, no fragments) paired with the extracted generalized CSS-selector template. See Section 4. Lumabyte (public template repository)
Website logs Basic Go server logs (IP address, user agent, request path, timestamp) kept for operational and abuse-prevention purposes. The lumabyte.com website loads no third-party analytics or crash-reporting scripts; there is no PostHog, no Sentry, no Google Analytics, and no advertising trackers on the site. Lumabyte

If you email us, or submit a form on the site (for example a waitlist or feedback form), we also hold the email address and whatever you chose to write. That data is used to reply to you and for product planning only.

Turning the check-in off

The version check-in is on by default in a packaged (non-development) build. It is switched off when:

  • You run LumaBrowser in development mode (unpackaged, or launched with --dev).
  • You block outbound traffic to lumabyte.com/api/pulse at the network layer, for example with a firewall rule.

Blocking the check-in does not disable any feature. Local chat, image generation, and everything else run fully offline; the app never needs to reach lumabyte.com after the models are downloaded, and it does not require "good standing" or any periodic contact to keep working.

4. Page templates, in detail

With analytics and crash reporting gone from the app, the shared template repository is the one genuinely novel privacy surface left in LumaBrowser, so it deserves a full explanation.

What it is

When you use the Template Builder, LumaBrowser extracts generalized CSS selectors from the page you're looking at (selectors that identify structural elements like "the primary navigation" or "the product price," not the actual text or values on the page). That selector template is sent to lumabyte.com and stored in a public, shared repository, the same repository that other users query when they land on the same kind of page. If you never use the Template Builder, nothing in this section applies to you.

Exactly what is transmitted

  • URL host - e.g. example.com
  • URL path - e.g. /products/item
  • The selector template - a JSON structure of CSS selectors

What is stripped before transmission

  • Query strings and URL fragments (so no ?token=, no #user-123)
  • Page content, form values, and any user-typed input
  • Cookies, local storage, session tokens
  • Your IP address and any personal identifiers

Why this matters and why we still do it

Domain + path is usually not sensitive for public web pages (e.g., amazon.com/gp/your-account). It can be sensitive for private or intranet URLs. If you work with URLs that reveal proprietary paths, internal tooling hostnames, or anything you wouldn't want in a public database, do not run the Template Builder on those pages, or block lumabyte.com at the network layer while you do; template generation itself keeps working locally.

We do it because crowdsourced templates are what make automation against the modern, randomized-CSS web tractable for everyone. The tradeoff is explicit, and sharing only happens when you use that one feature.

How to see or remove your templates

Templates are keyed by domain + path, not by user. Because there's no user identifier attached, we can't locate "your" templates on request. You can, however, request that templates for a specific host + path combination be removed from the public repository; email us at the address in Section 12 with the exact URL pattern.

5. How we use your data

We use the data in Section 3 for a short and specific list of purposes:

  • Delivering the service - serving installer downloads and updates, and operating the template repository.
  • Keeping the software working - understanding how many installs are active and which versions and platforms are in use so we can plan releases and support.
  • Abuse prevention and security - detecting attempts to overwhelm our infrastructure or misuse the template repository.
  • Replying to you - answering email and form submissions you send us.
  • Legal compliance - responding to lawful requests and enforcing our Terms of Service.

We do not sell your personal data, and we do not share it with advertisers.

6. Third parties we rely on

Very few. Because there is no account, no payment, and no license, we have no payment processor and no licensing service handling your data. The website and the check-in endpoint run on infrastructure we operate; the installer files are served from lumabyte.com. Model files you download come directly from the model hosts the app names in Setup (for example Hugging Face), under their terms, not through us.

We previously used PostHog (product analytics) and Sentry (crash reporting). Both have been removed entirely, from the LumaBrowser application and from the lumabyte.com website. We no longer send data to either. Earlier versions of this policy also described Stripe (payments) and Keygen.sh (licensing) for planned paid tiers; those tiers do not exist and neither service receives data from LumaBrowser. If we ever introduce a paid plan, this policy will be updated before it launches.

LLM providers you configure (Anthropic, OpenAI, local models, etc.) are not our processors; they are your processors. You bring your own API keys and your prompts go directly from your LumaBrowser instance to them. Their privacy terms govern that relationship.

7. Data retention

  • App version check-in - check-in records (version, platform, machine identifier, last-seen time) are kept only as long as needed for active-install counts and version and platform trend analysis.
  • Template repository - retained indefinitely, since it's a shared resource. Entries can be removed on request.
  • Server logs - rotated and purged on a short timeline (typically 30 days) except where retained for security investigation.
  • Email and form submissions - kept as long as needed to reply and plan, or removed on request.

8. Your rights

Depending on where you live, you may have some or all of the following rights over the personal data we hold about you:

  • Access - ask what we have.
  • Correction - ask us to fix inaccuracies.
  • Deletion - ask us to erase your data (subject to legal retention obligations).
  • Portability - get a copy in a portable format.
  • Objection / restriction - object to or restrict certain processing.
  • Withdraw consent - where processing is based on consent.

To exercise any of these, email us (see Section 12). We'll respond within a reasonable time, typically 30 days. If you're in the European Economic Area, the UK, or a US state with a comprehensive privacy law (e.g., CCPA/CPRA in California, VCDPA in Virginia, CPA in Colorado), you also have the right to lodge a complaint with your local supervisory authority.

Do Not Sell / Do Not Share (California): We do not sell personal information and do not share it for cross-context behavioral advertising. There is nothing to opt out of on that front.

9. Children's data

LumaBrowser is a general-purpose tool for adults and is not directed at children under 13 (or under 16 in jurisdictions where that's the applicable age). Some optional features (roleplay, image generation) can produce mature content and are intended for adults only. We do not knowingly collect personal data from children. If you believe a child has submitted information to us, contact us and we'll delete it.

10. International transfers

Lumabyte, LLC operates from the United States. Our service providers are also primarily US-based. If you use LumaBrowser or lumabyte.com from outside the US, your data will be transferred to, processed in, and stored in the United States. Where applicable, we rely on Standard Contractual Clauses or equivalent safeguards for transfers from the EEA, UK, or Switzerland.

11. Changes to this policy

We'll update this page when our practices change. The "Last updated" date at the top always reflects the most recent revision. For material changes, we'll post a notice on the site before the change takes effect.

12. Contact

For any privacy question, data-rights request, or template-removal request:

Lumabyte, LLC
Email: [email protected]
Web: lumabyte.com

For contractual terms, see /terms.