Imprint & privacy policy

Imprint

Sebastian Wiese-Wagner
Albert-Einstein-Straße 47
c/o Autorenglück #75241
02977 Hoyerswerda

Contact

Phone: +491753644013
E-Mail: seb@center.for.cybernetic.beauty

ATMOSYNTH is built with other people's work; their licenses and notices are reproduced in full.

Privacy policy

This Privacy Policy explains how Atmosynth ("we", "us", or "our") collects, uses, and shares information in connection with the Atmosynth service. If you have any questions, contact us at seb@center.for.cybernetic.beauty.

About Atmosynth

Atmosynth is a platform for creating, publishing and playing synth patches in the browser, built on top of the AT Protocol (ATProto), the same protocol used by Bluesky. ATProto is designed so that your data lives on your own Personal Data Server (PDS). Patches, modules and the samples they use are records and blobs in your own account, not ours — but some things are necessarily kept on our servers, and this policy says which.

Everything written to your PDS is public. An ATProto repository has no private records: anything Atmosynth publishes into your account on your behalf — a patch, a module, a like, an audio sample you upload — can be read by anyone on the network, through Atmosynth or through any other ATProto client, relay or index, whether or not they have ever visited this site. That includes everything a record carries: its name, description, tags, when it was made and changed, which modules and samples it uses and whose they are, which patch it was forked from, and, for a like, which patch you liked. All of it is published under your account identifier and handle. The things below that never reach any PDS stay private: a draft you have not published, the shared draft of a patch edited together, and who its collaborators are — the published record names nobody but its creator, the list is shown to the creator alone, and each collaborator sees only who else is editing while they are in it.

Information We Collect

• Cookies for authentication: we set two cookies, and only after you begin signing in. “atmo_sign_in” is set when you start a sign-in and lasts fifteen minutes: it holds a random value that names nobody, and exists strictly so that a sign-in can only be finished in the browser that started it. “atmo_session” is set once you are signed in: it holds an opaque session identifier and nothing else, and exists strictly to keep you signed in. We set no cookie before you start signing in, and we use no cookies for advertising, analytics or cross-site tracking. Separately, the player remembers how you like to play — volume, mute, the latency choice, the tempo, the arpeggiator, glide, unison and fixed-velocity settings, the output trim used when editing together — and the editor remembers whether notes snap to the grid, all in your own browser's storage. Those never reach our servers, and nothing about what you played or opened is written there.
• Your session: while you are signed in we store, on our servers, your account's identifier (DID) and handle, your display name and the address of your avatar, the access and refresh tokens that let us write to your account on your behalf, and the times the session was created and last used. Tokens stay on our servers and are never given to your browser. Signing out deletes that browser's session. A sign-in you begin and never finish leaves a short-lived record that expires after fifteen minutes.
• Drafts you have not published: work in the editor is autosaved to our servers, not to your PDS. A draft is private to you, has no public address and appears in no listing, and it is the one thing here that is not yet in your own account: it will not travel with you if you move your account until you publish it. We keep drafts until you publish or discard them. Audio you add to a draft is uploaded into your own account when you add it — the one act in an unpublished draft that writes to your PDS — but your PDS does not serve it to anyone, and deletes it after about an hour, unless a published record uses it. So we also keep a copy of it with your draft, play it from there, and upload it to your account again when you publish. The copy has no address anyone could find without your draft or a patch you edit together. We delete it when you publish what uses it, unless a draft or a shared draft still uses it, and otherwise a day after none does.
• Editing together: the creator of a published patch may name up to eight other accounts who can edit it with them. We store that list on our servers — the patch, each named account's identifier (DID) and when it was added. You enter a handle; we look up the account it names at that moment and keep the identifier, so a handle that later changes hands does not carry the patch with it. Nothing is written to anybody's account: the list is this site's own record of what the creator asked, like a draft, and the people on it are shown by handle and picture on the creator's own patch. While a patch has collaborators, its unpublished editing state is held on our servers as a shared draft — in memory while anybody has it open, in our database in between — instead of a private draft per person. It has no public address, appears in no listing, and is visible only to the creator and the accounts they named; everybody else sees the published record. Only the creator can publish it, and only into their own account. While you have such a patch open, the others editing it are shown who you are — your identifier, handle and picture — together with which block you have selected and which loops you have open. That presence, and where the patch's loops are in time, is held in our server's memory only, is never written down, and is gone when you close the page. The messages your browser exchanges to keep in time with the others carry timestamps and nothing about you. Removing a collaborator takes effect at once; removing the last one hands the shared draft back to the creator as an ordinary private draft (if it differs from the published record) and forgets the shared copy. Deleting the patch forgets both the list and the shared draft.
• Public ATProto data: we index and may cache data that is already public on ATProto — the names, descriptions, tags and timestamps of published patches and modules, who published them, which patches fork or use which modules, and which accounts have published a like of which patch — so that pages and search results can be listed without reading every account. We also cache the handle that belongs to an account identifier, and the fact that we have looked one up. All of this is derived from public records and can be rebuilt from them. We do not collect or store private ATProto data.
• Reports: if you report a patch we store what was reported, the handle of the account that published it, the reason you give, and the time. We do not store anything identifying the reporter — unless you put something identifying in the reason itself, which is a free-text field in your own words and is stored as you wrote it. To stop the report form being flooded we briefly note the coarsened network your request came from — the first half of the address, never the address itself — and discard it within a day.
• Moderation decisions: where we withhold a patch, a module or an account from this site, we keep a record of what was withheld, the reason, and when it was decided or lifted.

How We Use Information

We use the session cookie and the session it names to keep you signed in and to publish, edit and delete records in your account when you ask us to. We use drafts to give you back the work you left unfinished. We use the list of collaborators to decide who may open a patch's shared draft — checked when you connect and again every few seconds while you are in it — and the shared draft and presence to show everybody editing a patch the same state and what each of them has selected. We use the index of public data to render listings, search and creator pages. We use reports and moderation records to run the service. We do not profile you, sell anything about you, or share any of it with third parties for their own purposes.

Third Parties

We run no analytics and embed no third-party scripts or fonts. Two things do reach beyond this site: signing in sends you to your own PDS to authorise it, and an avatar is served by redirecting your browser to wherever that account publishes its picture, so that host — often a Bluesky CDN — sees the request. While you type a handle into the sign-in dialog, our server looks up matching accounts at typeahead.waow.tech, a community-run account search, and shows you the results. The service receives what you have typed so far, from our server rather than from your browser, so it does not see your address or anything else about you. Audio samples are streamed through our servers rather than linked, so playing a patch tells no one else that you did.

Data Storage and Retention

Your published content lives on your PDS, publicly, for as long as you keep it there. On our servers: drafts are kept until you publish or discard them, and the audio you add to them until it is published or a day after nothing unpublished uses it; a patch's list of collaborators and its shared draft are kept until the creator removes the last collaborator or deletes the patch, and who is present in it is not kept at all; a session is deleted when you sign out, and in any case ninety days after it was last used; unfinished sign-ins expire within fifteen minutes; cached handle lookups are dropped after a week; and the index of public data is kept as long as the record it describes is published.

Server Logs

The server writes a log line for every request: the method, the address requested, the response status and how long it took. Because the address is recorded in full, those lines show which pages, patches and samples were requested, and — since a search term travels in the address — what was searched for. They do not include request headers, so no cookie, no session identifier and no browser identification is written, and Atmosynth itself writes no IP address to them. Opening a patch's shared draft is one such request, so the log shows that a connection to that patch was opened; the edits and presence exchanged over it afterwards are not logged.

The log goes to the server's standard output. Atmosynth keeps no log file of its own and applies no retention to it, so how long those lines survive is a property of how this deployment is hosted rather than of the software. Any reverse proxy or hosting provider in front of Atmosynth keeps its own logs, and those ordinarily do record IP addresses.

Your Choices and Deletion

Because your published data lives on your PDS, you can remove your content at the source using Atmosynth or any other ATProto client, and it stops being listed here. Because it was public while it was there, other services on the network may have indexed or copied it in the meantime, and removing it from your account does not reach those copies. You can discard any draft from the editor, and sign out to delete a session. As a patch's creator you can remove any collaborator, or all of them, from the editor's menu at any time. If you have been named as a collaborator on somebody else's patch and would rather not be, ask them to remove you, or write to us. If you would like us to delete anything else we hold on our servers, write to seb@center.for.cybernetic.beauty.

Children’s Privacy

Atmosynth is not directed to children under 13, and we do not knowingly collect personal information from children.

Changes to This Policy

We may update this Privacy Policy from time to time. When we do, we will revise the “Last updated” date below. Your continued use of Atmosynth after an update means you accept the changes.

Last updated: 2026-09-20