What the app collects
Three things. The first two are on by default and can each be switched off in Settings, under General.
- Crash reports. If the app crashes, a report is sent so the fault can be fixed. It contains the stack trace, the Hydra version and your macOS version. It carries no connection details, hostnames, credentials or file contents, and it is not tied to you or to your machine. Handled by Bugsnag. Turning this off takes effect the next time Hydra launches, because a crash reporter cannot be safely half-removed once it is running.
- Usage analytics. Which features get used, plus the Hydra and macOS version, device model, CPU architecture, screen size, language, timezone and country. These are tied to a random identifier generated on first launch, which stays the same for that installation. Our server also records the IP address each request arrives from, as any web server does. Event names come from a fixed list in the app: they record that a feature was used, never what you used it on, so no connection name, hostname, path, query or file content is included. Received by our own code, running on a server we rent from Microsoft Azure; no third-party analytics product is involved, and Microsoft sees the traffic the way any hosting company sees its tenants' traffic. Turning this off stops it immediately.
- Your email address, if you give it. The app's welcome flow has an optional screen offering to keep you posted, and the download page has an optional email field that says the same thing. If you leave it blank, nothing is collected. If you fill it in, the address is used for exactly what that screen said: in current versions, a short getting-started series over your first two weeks and then occasional release notes; if you signed up on version 2026.274 or earlier, release notes only. Every email has an unsubscribe link, the mail is sent through Resend, and the address is never sold or shared for anyone else's use.
A random per-installation identifier alongside an IP address is not the same as anonymous data, and we would rather say so plainly than claim otherwise. If you would prefer not to send it, the switch is in Settings, under General, and it works immediately.
What it does not collect
- No advertising identifiers, and nothing shared with or sold to advertisers or data brokers.
- No record of what you connect to, open, edit or transfer. Connection names, hostnames, file paths, queries and folder contents are never sent.
- No copies of your files, queries, database contents or credentials.
- No account. The app does not require you to sign up for anything.
Where your connection library lives
On your Mac. If you are signed in to iCloud, it also syncs through your own private iCloud database, which is your Apple account rather than a server operated by Hankinsoft. If iCloud is unavailable the app runs entirely locally and says so in its status bar.
Where your passwords live
In the Keychain, stored separately from the connection library. Passwords, access keys and API tokens are marked for iCloud Keychain sync, so they follow you to your other Macs the way Safari's passwords do; SSH private keys, and the tokens and certificate that identify one particular Mac, stay on the Mac they were created on. None of it is ever written into the synced library database, and none of it is ever transmitted to Hankinsoft.
Scanning a project folder
When you ask the app to scan a folder, the walk happens on your Mac. The judgement about whether a discovered credential is a real secret or a placeholder is made by a model running on your device. The contents of your configuration files are not uploaded anywhere for that, or for anything else. Nothing found in a scan is imported until you select it, and secret values are not displayed on screen.
Network connections the app makes
The app connects to the servers you configure. Everything else it contacts is listed here, because a list that leaves something out is worse than no list.
- Update check (direct-download edition only): hydradevstudio.com, for the appcast. Sends the app version.
- Crash reports and usage analytics, as above, while their switches are on.
- Licence check. When a licence key is entered, at launch, and when the licence pane is opened, the app asks our licence server whether the key's update window has been extended or refunded. It sends the key and the email address it was registered with. It sends nothing about your Mac: no serial number, hardware identifier or hostname. If the server cannot be reached the app stays exactly as licensed as it was.
- The Buy button opens our store in your browser with the app version and the number of trial days remaining in the link, so the page can say the right thing.
- No cloud AI. The one AI feature is optional: after scanning a project folder for a new workspace, Hydra offers to review what it found with Apple Intelligence. That runs on Apple's on-device model, on your Mac, and sends nothing anywhere. There are no AI provider settings and no API keys to enter.
- The hidden web page, only when an automation client asks for it. An MCP client on this Mac can open web pages inside Hydra and read them. It fetches the sites the client asks for, public sites only unless you allow local addresses in Settings, keeps its own cookies like a browser profile, and Settings shows how many sites it holds data for and can clear them.
Two optional automation surfaces, a local control socket and a network server, exist for scripting and are switched off by default. A program that connects to the socket is named and approved by you the first time, and everything automation does is recorded in an activity log. That log syncs with your library through your own iCloud database, so it is on your other Macs too; it holds connection names, paths and outcomes, and never a credential or a query's string values.
This website
hydradevstudio.com serves static pages, and uses Google Analytics to count visits and see which pages get read. That is a third-party service: it sets its own cookies in your browser, and your IP address, the pages you view and rough location go to Google. No advertising or remarketing features are switched on, nothing here is sold to data brokers, and none of it is joined to anything the app collects.
This is worth saying plainly, because the app deliberately does the opposite: its usage analytics go to our own code on a server we rent, precisely so that no analytics company is involved. The website is the one place that is not true, and a content blocker will stop it. Nothing on the site needs it to work.
Contact
Questions about any of the above can go to support.
Last updated 16 September 2026.