FileFluxFileFlux — Privacy Policy
Last updated: 2026-10-20
FileFlux is a download manager that runs entirely on your own PC (Windows or Linux). It has no server, collects no analytics or telemetry, and sends none of your data to the developer. This policy explains what data FileFlux handles, where it stays, and what is sent elsewhere when you choose to use an optional feature that needs it.
In short
- Your data stays on your device. The one exception is a hosted AI service you choose to set up: when you use an AI feature with it, the details that feature needs are sent to that service (see Optional AI features).
- FileFlux connects to the internet only to download the files you ask it to. If you turn them on, it also connects to the proxy and encrypted-DNS servers you choose, and to the AI service you set up.
- AI features are off until you turn them on, and are never used in Incognito mode.
- No accounts, no tracking, no advertising, no selling or sharing of personal data.
What FileFlux stores on your device
- Download history (URLs, file names, sizes, status, dates) — saved locally and encrypted at rest, with Windows DPAPI on Windows and your desktop keyring (the freedesktop Secret Service) on Linux. You can clear it on exit, or use Incognito mode to keep downloads in memory only so nothing is written to disk.
- Settings (download folder, speed limits, proxy/DNS preferences, etc.) — saved locally. Proxy passwords and AI service API keys are encrypted the same way.
- Saved site logins, feeds and saved views — saved locally and encrypted at rest.
- Resume data for in-progress downloads — saved locally (encrypted) so interrupted downloads can continue where they left off.
- The source of each downloaded file — like a web browser, FileFlux marks a finished file with the address it came from (on Windows, the "Mark of the Web" that lets Windows warn you before running a downloaded program). Passwords and query strings are left out of that address. For Incognito downloads the warning mark is kept but no address is recorded.
- Agent activity log — only if you let AI agents control FileFlux (see AI agents). A local record of what an agent asked for, with web addresses shortened to leave out query strings and passwords. Nothing is recorded in Incognito mode. You can turn it off or clear it.
- Diagnostic log — a small local file of startup timings and error details, kept to help fix problems. Your account name, web addresses and site names are removed from it. It is never sent anywhere; you can attach it to a bug report if you choose to.
Data from the FileFlux browser extension (optional)
If you install the companion browser extension, it passes a download's URL, the page referrer, your browser's User-Agent, and the cookies for that site to FileFlux over the local Native Messaging channel. This is used only to perform the download (for example, so downloads that require you to be signed in still work). It is sent only to the FileFlux app on the same computer and is never transmitted to the developer or any third party. FileFlux does not record which browser a download came from.
Optional AI features
AI features (tag suggestions, renaming, summaries, diagnosing failed downloads, finding mirrors, extracting links and checking links) are off until you turn them on in Settings → AI, and are never used in Incognito mode.
Where your data goes depends on the AI service you choose:
- On this computer — the on-device model built into Windows (where available), or a server you run yourself such as Ollama or LM Studio. Nothing leaves your computer.
- On your local network — a server you run on another machine you control. Nothing leaves your network.
- A hosted service — Anthropic (Claude), OpenAI or another OpenAI-compatible service, or Google (Gemini). What a feature needs is sent to that company and handled under its privacy policy and terms. FileFlux sends it nowhere else, and the developer never sees it.
What is sent, and when. Only the details the feature you are using needs, at the moment it runs:
- Suggest tags, rename, organise into folders: the file name; the web address it came from, shortened (see below); your existing tag or folder names. Sent when you use the feature, or automatically for each new download if you turn on Auto-tag.
- Analyse the file for tags (an optional extra to Auto-tag): up to about 2,000 characters of text from the finished file. Sent automatically when a download finishes, only if you turn it on.
- Summarise: the file name and an excerpt of the file's text. Sent when you ask for a summary.
- Diagnose a failed download, find mirrors: the file name and size; the shortened web address; technical details of what went wrong (for example error codes and retry counts). Sent when you ask.
- Extract links: the text you paste, exactly as you pasted it. Sent when you use Extract links with AI.
- AI link check (optional): the shortened web address, and the file type, name and size the server reported. Sent automatically while links are checked before downloading, only if you turn it on.
- Connect: a request for the list of available models. Sent when you click Connect.
Shortened web addresses. Before a web address is sent to an AI service, FileFlux removes the parts that commonly carry passwords or access tokens: the user name and password, everything after "?" or "#", and any part of the path that looks like a secret key rather than a name. Cookies, browser headers and saved logins are never sent to an AI service.
API keys you enter are stored encrypted and sent only to the address of the service you configured. FileFlux refuses to send a key over an unencrypted (http) connection to an address on the internet.
Proxy. By default, AI requests connect directly. If you turn on Send AI requests through the proxy, requests to hosted services use the proxy and DNS settings your downloads use. Servers on your own computer or local network are always reached directly.
Mirror suggestions. When you ask FileFlux to find another source for a download, an AI service may suggest web addresses. FileFlux then contacts those addresses itself to check that they serve the same file before using them. It only contacts secure (https) addresses on the public internet, and sends them none of your cookies or logins.
AI agents (optional)
If you turn on Agent control, AI assistants running on your computer (through FileFlux's MCP server) can see your download list — file names, web addresses and progress — and can add, pause, resume, queue or re-prioritise downloads, within the permissions you set. They cannot delete files or change settings. This connection stays on your computer, and agents get no answer in Incognito mode. However, the assistant itself may send what it reads to its own AI service, under that assistant's privacy policy.
Network connections
FileFlux makes network connections only:
- to the websites and servers you choose to download from (HTTP/HTTPS/FTP), and for torrents, to the relevant trackers and peers;
- to other sources of the same file — mirrors that a download's server advertises, ones you add, and, when you ask, ones an AI service suggests (checked as described above);
- to the RSS/Atom feeds you subscribe to;
- to a proxy server (HTTP or SOCKS5/Tor) if you configure one;
- to a DNS provider (Cloudflare, Google, Quad9, or one you specify) if you enable encrypted DNS (DoH/DoT);
- to the AI service you set up, when you use an AI feature (see Optional AI features);
- when you click Check for updates: the Microsoft Store (Windows) or GitHub's release listing for FileFlux (Linux).
It contacts no other servers and performs no background reporting.
What FileFlux does NOT do
- No analytics, telemetry, crash reporting to the developer, or usage tracking.
- No advertising, and no sale or sharing of personal data.
- No user accounts and no cloud storage.
- Never sends your data to the developer, including through the AI features.
Your control
You can delete your download history at any time (and have it cleared automatically on exit), use Incognito mode for trace-free downloads, turn AI features and agent control on or off at any time, and uninstalling FileFlux removes its local data.
Children's privacy
FileFlux is a general-purpose utility, is not directed at children, and does not knowingly collect personal information from anyone.
Changes to this policy
This policy may be updated from time to time; the "last updated" date above will change accordingly.
Contact
Questions about this policy: support@fileflux.online