capy Privacy Policy
This Privacy Policy explains how the capy application accesses, uses, stores and shares your data, including the Google user data capy can access after you sign in with Google and grant consent, and how you can revoke access and delete your data.
The short version: capy is an open-source program that runs on your own computer. It has no backend server. Your data stays on your computer, in your own Google Drive, and on the music services you connect. No user data or credentials ever pass through us, and we have no means to collect them.
1. What this policy covers
This policy applies to the capy application (the command-line tool and its built-in local web interface, capy --web) and to this website, capy.taislife.work. capy is maintained by Tai (GitHub: Tai-ch0802) as an individual. The source code is public on GitHub under the MIT License, so every statement below can be verified in the code.
2. Google user data capy accesses
Signing in with Google is optional. When you run capy auth login google and grant consent, capy requests exactly the following three scopes and no others:
openid: identifies which Google account signed in.https://www.googleapis.com/auth/userinfo.email: the email address of that account.https://www.googleapis.com/auth/drive.appdata: read and write files that capy itself creates in the application data folder (appDataFolder) of your Google Drive. This scope cannot read, list or modify any other file in your Drive.
3. How that data is used
- Your email address is used only to show, on capy's own screens, which account is signed in (
capy auth statusand the web UI's account page, when sign-in completes, and in the confirmations and errors shown before capy writes or restores data on Drive), so that you can notice if you signed in to the wrong account. capy never emails you and does not use the address for anything else. - The application data folder stores what is needed to keep playlists in sync across your devices: playlists (name, description, tracks and their order, and which service playlists they are linked to), track data (title, artist, album, duration, ISRC, and the track's ID on each music service; if you connect a music folder on your computer, that "ID" is usually the file's path relative to that folder, but if the playlist file (M3U) lists an absolute path or a path outside that folder, it is recorded as written, which can include your computer's account name and folder structure; a local playlist's ID is "device ID/playlist file name", so paths and file names you chose yourself may be included. If a local music file has no entry in
library.jsonand the playlist file gives it no#EXTINFtitle, capy uses its file name (without the extension) as the track's title, and a local playlist's name is its file name, so the "title" and "playlist name" sent in the items below may be file names you chose), a device list (a random ID capy generates for each computer, the computer's name, and the last sync time), and a per-device snapshot of what each playlist looked like at the last sync. If you connect YouTube Music, the id of a linked YouTube Music playlist is "YouTube channel ID/playlist ID"; the channel ID is that YouTube account's public identifier (it starts with UC) and tells which account the playlist belongs to. - Syncing and moving: when you add a song to a master playlist, sync or move a playlist or look up matching tracks (for example
capy pl add/pull/push/sync/dedup,capy pl link --create/--merge,capy migrate,capy resolve, and the web UI buttons that do the same), capy takes the tracks' identifiers from the playlist and track data above (such as ISRC, title, artist and the track's ID on each music service), plus the playlist's name when it creates a playlist on a service or syncs a rename, and the linked playlist's ID on that service when it reads or writes that playlist (sent only back to the service that issued it), and sends them, at your direction, to the music services you are signed in to, to look up the same song or read and write the playlist (when you runcapy pl add --nowon macOS for a song playing in Apple Music, capy also looks up its title and artist in the Apple Music catalog to get its catalog ID); when it looks for the same song on YouTube Music, the title and artist are sent to Google as search queries (see section 4). This happens only when you run these operations. - Playing from My playlists: when you press ▶ on a song or playlist (or 🔀 to shuffle-play it; the same applies when playing a master copy via
capy playor interactive mode's drawer) in the web UI's My playlists page, capy sends that song's or playlist's track IDs (or linked playlist ID) on that service to Spotify or Apple to play it (Apple first looks the ID up in its catalog); the "Listen on Spotify" and YouTube Music "Open" links on that page, the same links next to a linked playlist in its links panel and on the Search page, and the Spotify links in change tables and in the move wizard's list of songs that could not be moved carry that song's or playlist's ID on that service and are opened by your browser only when you click them. This happens only when you press it. - The song wiki's Wiki button: when you press Wiki on a song in the web UI's My playlists page, capy takes that song's title and artist from the track data above and sends them, together with the native language you set, to the AI endpoint you configured yourself with
capy wiki setup, to write the song wiki you asked for (see sections 4 and 5). This happens only when you press the button; nothing else from your Drive is sent. The answer ends with an MV link: a YouTube search URL made of that same title and artist, which your browser opens only if you click it. - Showing it to you on your computer: the web interface's (
capy --web) My playlists and ISRC lookup pages display the playlist and track data above, andcapy exportprints all of it, at your direction, for you to back up; all of this stays on your computer and is not sent anywhere. - capy does not use Google user data for advertising, does not use it to train any AI or machine-learning model, does not build user profiles, and does not use it for any purpose other than those described in this section.
4. Where data is stored
- Your Google Drive (application data folder): the sync data listed in section 3. It counts against your Drive storage, is not visible in the Drive interface, and cannot be shared with anyone.
- Your computer: a local cache (an SQLite file) holding a copy of the same sync data, plus your music services' playlist lists, the playlist tracks read during sync (those tracks are reused for at most a week), unmatched track-matching results cached for at most a week, your last 50 searches and played items, your last 1,000 listening entries recorded after 30 seconds of playback, and your last 100 interactive command history entries (clear all search, playback, listening and command history with
capy history clear), and song wiki answers (title, artist, language, model and the answer text); a settings file (which includes the email address above, and the name, handle and channel ID of the YouTube Music account you logged in to); and access tokens for Google, Spotify and Apple Music, the YouTube Music login cookie (only the three session cookies), plus the API key and custom headers of an AI endpoint you set up yourself, which are stored only in the operating system keychain (macOS Keychain or Windows Credential Manager), never in settings files, the database, or Drive. - On our side: nothing. capy has no backend server, no account system, and no telemetry or crash reporting. The program connects only to the services you use (Google, Spotify, Apple, YouTube Music). YouTube Music is connected with a login cookie you copy yourself from music.youtube.com (not officially supported by Google); capy uses it to read and write the playlists in your YouTube Music account. When
capy migrate,capy resolveorcapy pl link --mergelooks for the same song on YouTube Music, it sends the song titles and artists from your playlists to Google as search queries; when acapy --webpage shows album art or plays a preview, your browser loads those images and audio files directly from Spotify's or Apple's content servers. Apart from that, capy contacts GitHub only when you runcapy update, to check for and download a new version (capy update --devinstead has the Go toolchain on your computer download the source code from GitHub and the Go module proxy). If you connect your own AI endpoint withcapy wiki setup(capy has no default endpoint),capy wiki setupcontacts it to check the connection and list its models; after that, only when you runcapy wiki, capy sends that song's title, artist, album, release date and genre (only those the source actually provides: when you ask by title, just the title and artist you typed; for the song that's playing, just what the platform returns), plus the native language you set (used only to choose the answer's language), to that endpoint — a service you chose (for example OpenAI, or one you host yourself), under its own policy. The MV link in the song wiki is just a YouTube search URL that your browser opens; the song wiki itself never contacts YouTube.
5. Sharing and transfer
capy does not sell your Google user data and does not give it to us, and it does not share or disclose it to any third party except in the cases below, which you trigger yourself. It sends the data, as described in section 3, only in the cases below, only when you trigger it, only for the feature you are using, and only to services you signed in to or configured yourself: to the music services you are signed in to when you sync, move or look up matching tracks (and to Google as search queries when looking for a song on YouTube Music), to that music service to play a song when you press ▶ on My playlists, and to the AI endpoint you configured when you press Wiki. In addition, when you click a link on the web UI (such as "Listen on Spotify", YouTube Music's "Open", or the song wiki's MV link), your browser opens that page, and the link carries that song's or playlist's ID on that service (for the MV link, the song's title and artist as a YouTube search). Apart from that, the data never leaves your computer and your own Google account. capy never uses this data to train any AI or machine-learning model; how a service that receives it handles it is governed by that service's own policy.
When the song wiki (capy wiki) is about the song that's playing or a title you type, it sends the public song details listed in the previous section to the AI endpoint you configured yourself, at your direction; those come from the music service or from what you typed, so they are data about the music, not your Google user data, and capy sends no account information with them.
The Wiki button on My playlists is different: the title and artist it sends are taken from your playlist's master copy stored in your Google Drive, so they are Google user data. capy transfers them only to provide the song wiki you asked to see, only when you press the button, and only to the endpoint you configured, and never uses them to train any AI or machine-learning model. Whether that endpoint keeps what it receives or learns from it is governed by that service's own policy, so choose one whose terms you accept.
capy's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
6. Retention and deletion
Because the data is in your hands, you decide how long it is kept. To remove everything:
- Revoke access: remove capy under Google Account → Third-party apps and services.
- Delete the data in Drive: Google Drive → Settings → Manage apps → capy → "Delete hidden app data".
- Delete the data on your computer: run
capy auth logout google(and, for the other platforms you connected,capy auth logout spotify,capy auth logout appleandcapy auth logout youtube; the last one deletes the YouTube Music login cookie and the account recorded in the settings file) to clear the tokens from the keychain. If you set up an AI endpoint for the song wiki, its API key and custom headers are in the keychain too, and no logout command removes them: in Keychain Access (macOS), delete thecapy-musicitems whose account isai.api_keyorai.headers; in Credential Manager (Windows), delete the generic credentialscapy-music:ai.api_keyandcapy-music:ai.headers. Then delete capy's settings folder (macOS:~/Library/Application Support/capy-music; Windows:%AppData%\capy-music). If you are worried that the YouTube Music login cookie has leaked, sign out of all devices in your Google Account → Security and it stops working immediately.
7. Security
All connections between capy and Google, Spotify, Apple and YouTube Music use HTTPS. Tokens are stored only in the operating system keychain. The capy --web interface binds only to 127.0.0.1 (reachable only from your own computer), uses a one-time access code on every start, and never shows any token or secret on the page. If the AI endpoint you set up uses plain http to another computer, capy warns you when you set it (the API key and custom headers would travel unencrypted).
8. This website
capy.taislife.work is a static website: it sets no cookies, loads no third-party resources, has no forms, and contains no analytics or tracking code of our own. It is hosted by Cloudflare, and the hosting provider may automatically add analytics code to pages at the domain level. This site's Content Security Policy (default-src 'none') forbids pages from running or loading any script; in our tests that code is blocked by your browser and is neither executed nor loaded.
As with any website, Cloudflare processes the technical information needed to serve a request (such as IP addresses and the headers your browser sends with the request) under its own privacy policy. Our Cloudflare account may keep access logs (the requested URL, the time, and the technical information above) for a limited period, solely to troubleshoot the website; we do not use those records to identify or track visitors. This site shares its parent domain with taislife.work: if you have visited other pages on taislife.work, your browser may include cookies set by those pages in the request headers it sends to this subdomain. Like any other header they reach the hosting provider and are covered by the previous sentence; nothing on this website reads or uses them, and we never look at them.
9. Children
capy is not directed to children under 13, and we do not (and cannot) collect personal data from anyone.
10. Changes to this policy
If the way capy accesses data changes (for example, a new permission), we will update this page and the effective date above before releasing the new version. Every revision is kept in this file's version history.
11. Contact
Questions about this policy or your data: please open an issue on GitHub Issues (do not paste tokens or personal data into a public issue).