CLI tool
iterate-ch/cyberduck avatar
iterate-ch/cyberduck

Cyberduck: a GPL file transfer client for S3, SFTP and WebDAV

Cyberduck is a libre FTP, SFTP, WebDAV, Amazon S3, Backblaze B2, Microsoft Azure & OneDrive and OpenStack Swift file transfer client for Mac and Windows.

4,912 stars393 forksJavaGPL-3.0

At a glance

What is it?
Cyberduck is a libre FTP, SFTP, WebDAV, S3, B2, Azure and OpenStack Swift client for macOS and Windows, with a separate CLI for Linux, macOS and Windows. The desktop app is the product; the repository is a Maven and Ant build that also produces installers and a headless duck binary.
Who is it for?
Install Cyberduck if you move files between FTP, SFTP, WebDAV and object storage from a desktop and want one client with a bookmark list instead of a separate tool per protocol. Do not adopt it if you need a scriptable transfer engine inside a Linux service: the CLI is a separate artifact from the desktop app, and the Docker image in the repository only carries the duck binary.
Can I use it commercially?
Yes, with conditions. GPL-3.0 is a copyleft licence: if you distribute software that includes it, you must release that software's source code under the same licence. Running it internally without distributing it does not trigger that obligation.
Is it still maintained?
Yes. The repository last received commits 2 days ago.
What is it written in?
Mainly Java, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • iterate-ch/cyberduck4,912 ★ · +109
Star history
Repository2026-09-232026-11-15
iterate-ch/cyberduck4.8k4.9k

109 Stars in the last 53 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

What Cyberduck is for, and who ends up using it

Cyberduck is a desktop file transfer client. The README describes it as a "libre file transfer client for macOS and Windows" and lists the protocols in the repository description: FTP, SFTP, WebDAV, Amazon S3, Backblaze B2, Microsoft Azure and OneDrive, and OpenStack Swift. That is a wide surface for a single application, and it is the reason people reach for it: one bookmark list, one transfer queue, one set of preferences, instead of an FTP client plus an S3 browser plus something for WebDAV.

The audience is fairly specific. If you administer a web host over SFTP, move objects into an S3 bucket, or mount a WebDAV share on a corporate network, Cyberduck covers all three from the same window. The top-level directories go further than the description, with folders such as box/, brick/, cryptomator/, ctera/, deepbox/, dracoon/ and dropbox/.

Two things about the project shape matter before you install anything. First, the README says the core libraries are used in Mountain Duck, a separate product. That means the transfer engine is shared with something sold elsewhere, which is a good sign for the engine's maturity and a reason to expect the open source app and the commercial app to diverge in interface rather than in protocol support. Second, the README points to a separate repository for "Additional Connection Profiles" that are not bundled by default and appear under Preferences, Profiles. If the service you care about is not in the default list, that is where it lives.

How the repository is laid out, and what that tells you about the build

The top level of the repository is a multi-module build, and the directory names are the architecture. Each protocol or backend has its own directory: azure/, backblaze/, box/, brick/, ctera/, cryptomator/, deepbox/, dracoon/, dropbox/. Shared code sits in core/ and binding/. The user interfaces are separate: osx/ and windows/ for the desktop apps, cli/ for the command line client.

That split explains the most important structural fact about Cyberduck: the CLI is not a mode of the desktop application. It is its own artifact, built from cli/ and released for macOS, Windows and Linux. The README's build section confirms this by listing installer outputs for cli/osx, cli/windows and cli/linux alongside the desktop packages. The Dockerfile in the repository root follows the same pattern: it copies opt/duck, symlinks /opt/duck/bin/duck to /usr/bin/duck, and sets the entrypoint to /usr/bin/duck with --version as the default command. The image contains the duck binary, not the graphical client.

Build tooling is mixed by necessity. The prerequisites are Java 11 SDK or later, Apache Ant 1.10.1 or later, and Apache Maven 3.5 or later. On macOS, Xcode 12 or later is required. On Windows, the README asks for Visual Studio 2022 with the .NET desktop development, Universal Windows Platform development, and Desktop development with C++ workloads, plus the Bonjour SDK for Windows and optionally WiX v3. The Windows build also needs NuGet credentials for the GitHub Package Registry sources gh-ikvmnet and gh-iterate-ch, configured with a classic personal access token that has at least read:packages permission. That is a heavier toolchain than a typical Java project, and it is the first real cost of working from source rather than from a release.

Building Cyberduck from source on macOS, Windows or Linux

The README gives one command for a plain build without tests and without code signing. It is the shortest path to an artifact:

bash
SKIP_SIGN=true mvn verify -DskipTests

After it completes, the README says build artifacts land in osx/target/Cyberduck.app and windows/target/Cyberduck.exe. A build on a machine without the platform toolchains will not produce both; the macOS app needs Xcode, and the Windows executable needs the Visual Studio workloads listed in the prerequisites.

Add the installer profile when you want distributable packages rather than a runnable app directory:

bash
SKIP_SIGN=true mvn verify -DskipTests -Pinstaller

The README lists the outputs as osx/target/release/*.(zip|pkg), windows/target/release/*.(exe|msi), cli/osx/target/release/*.(pkg|tar.gz), cli/windows/target/release/*.(exe|msi) and cli/linux/target/release/*.(deb|rpm). If your interest is the headless client, that last pair is the one to look at.

For macOS sandbox builds, the README documents a profile that applies the com.apple.security.app-sandbox entitlement:

bash
SKIP_SIGN=true mvn verify -DskipTests -Psandbox

If you would rather not build at all, the Dockerfile gives a container whose entrypoint is the duck binary and whose default command prints the version. The image is a starting point you assemble yourself from the file as written. The same is true of day to day use: the README states that general help lives in the documentation repository at docs.cyberduck.io, and that is where connection setup, bookmarks and transfer options are described.

The Windows build is the part most likely to cost you a day

The README is unusually candid here, which is worth noting because most project READMEs are not. Under the Chocolatey installation path it writes: "Installing with Chocolatey may or may not fail spectacularly." It then lists observed failures on a clean installation. The Bonjour package fails with file not found even though Bonjour64.msi is extracted from BonjourPSSetup.exe. The wixtoolset package depends on a .NET 3.5 package that never completes, and the README says installation does not work on Windows 11. The visualstudio*-workload-* packages may halt with "Operation canceled", in which case the instruction is to abort the Chocolatey command with Ctrl-C and resume the installation from the Visual Studio Installer.

That is not a reason to avoid the project. It is a reason to plan the Windows build as a task with a fallback, not as a five minute step. The manual path exists: install Visual Studio 2022 with the three named workloads, the Bonjour SDK for Windows, and optionally WiX v3. Then confirm that MSBuild, mvn, ant and java are on PATH. The README suggests running where msbuild.exe from a Developer Command Prompt for VS2022 and adding the first directory to PATH, and notes that the Microsoft OpenJDK 17 installer adds itself to the system PATH. You also need the latest Windows SDK binary folder on PATH, in the form %ProgramFiles(x86)%\Windows Kits\10\bin\10.0.<Latest>.0\x64.

The NuGet step is easy to miss and will stop the build. Before building on Windows, configure credentials for the two GitHub Package Registry sources:

bash
dotnet nuget update source gh-ikvmnet -u "YourUsername" -p "YourPAT"
dotnet nuget update source gh-iterate-ch -u "YourUsername" -p "YourPAT"

Replace YourUsername with your GitHub username and YourPAT with a classic personal access token carrying at least read:packages. The README also says to restart the machine after installing the components.

Where Cyberduck is the wrong tool

The clearest limitation is the one the repository layout implies rather than states. Cyberduck is a desktop client first. The graphical application targets macOS and Windows only. Linux users get the CLI, built from cli/linux into .deb and .rpm packages, and the repository's own Dockerfile wraps the duck binary rather than a GUI. If your workflow assumes a Linux desktop with a file browser integration, this is not that product.

Scripting is the second boundary. The CLI exists and is packaged for all three platforms. Anyone planning to replace a shell script that drives sftp or aws s3 with duck should read the documentation at docs.cyberduck.io first.

Third, protocol coverage is not the same as protocol depth. The README explicitly separates bundled profiles from the additional connection profiles maintained in their own repository and reachable through Preferences, Profiles. A backend you need may still require that extra step before it shows up as a connection option. Verify the specific service before you standardize on the client.

Finally, the README warns about snapshot and beta builds: they are "nightly snapshot builds from the current development trunk" and "potentially unstable and experimental", switchable under Preferences, Update. That is a normal arrangement, but it means the update channel is a setting a user can change, and a support conversation about a strange failure should start by confirming which channel the machine is on.

Cyberduck compared with FileZilla and with command line transfer tools

The comparison people actually search for is Cyberduck versus FileZilla, and the difference is scope rather than quality. FileZilla is an FTP, FTPS and SFTP client. Cyberduck covers those protocols and adds WebDAV and a long list of object storage and cloud backends, from S3 and Backblaze B2 to Azure, OneDrive and OpenStack Swift. If your work stops at FTP and SFTP, the extra surface buys you nothing and costs you a larger application. If your work crosses between an SFTP host and an S3 bucket in the same afternoon, the single bookmark list is the whole point.

The second alternative is the command line itself: sftp, scp, rsync, or a cloud provider's own CLI. Those tools are scriptable by default and are the right answer when the transfer is part of a pipeline. Cyberduck's answer to that world is the duck binary, packaged for Linux, macOS and Windows, and the Docker image built from the repository's Dockerfile. The trade-off is documentation: the provider CLIs ship their own reference material, while the Cyberduck README points you to docs.cyberduck.io for anything beyond building.

A third option is Mountain Duck, which the README names as a consumer of the same core libraries. It is a different product, so treat any feature comparison as something to check on the respective sites rather than infer from this repository. What the README does establish is that the transfer engine underneath is shared, which is a reasonable signal about where protocol fixes land first.

Licence, maintenance and the cost of staying current

Cyberduck is licensed under the GNU General Public License Version 3. The README states the source is available under that licence and that pull requests are welcome. For an internal tool that is usually unproblematic. For redistribution, the GPL-3.0 obligations attach to the binary you ship, and that is a question for your own counsel rather than for this article. The practical point is narrower: if you build from source and hand the result to colleagues, you are distributing a GPL-3.0 work, and the licence text is in the repository as LICENSE.txt.

Maintenance looks current. The repository is not archived, and the last push was on 2026-11-01. The README documents snapshot and beta channels under Preferences, Update, and the localization workflow runs through Transifex with a mailing list for translation changes. There is a separate news mailing list for release announcements and a Google Group for discussion. None of that is a guarantee about release cadence, and the repository has no GitHub releases, so the honest statement is that the trunk is active and the release history is something to check in CHANGELOG.md or on the project site.

Upgrade cost depends on how you installed it. A build from source means re-running the Maven command and re-checking the Windows NuGet credentials and the platform toolchains each time. A packaged install means the update channel setting under Preferences, Update decides what you get, and a machine switched to snapshot builds will track the trunk. For a team, the decision worth making explicitly is which of those two you are on, because they fail in different ways: a broken trunk build versus a stale packaged release.

Editorial conclusion

Install Cyberduck if you move files between FTP, SFTP, WebDAV and object storage from a desktop and want one client with a bookmark list instead of a separate tool per protocol. Do not adopt it if you need a scriptable transfer engine inside a Linux service: the CLI is a separate artifact from the desktop app, and the Docker image in the repository only carries the duck binary. Before you commit, verify that the protocol you need is in the bundled set rather than in the separately maintained profiles repository, and check that your distribution route gives you a build whose update channel you control.

Frequently asked questions

What is Cyberduck used for?

It is a file transfer client for macOS and Windows that speaks FTP, SFTP, WebDAV, Amazon S3, Backblaze B2, Microsoft Azure and OneDrive, and OpenStack Swift, among others. A separate command line client, duck, is built for Linux, macOS and Windows.

How do I install Cyberduck on macOS?

The README points to docs.cyberduck.io for general help. From source, the documented command is SKIP_SIGN=true mvn verify -DskipTests, which produces osx/target/Cyberduck.app, and adding -Pinstaller produces packages under osx/target/release.

How do I use Cyberduck with Amazon S3?

S3 is one of the protocols the project lists. The README states that usage documentation lives in the separate docs repository at docs.cyberduck.io.

How do I use the Cyberduck CLI?

The CLI is a separate artifact built from cli/, with installer outputs for macOS, Windows and Linux, and the repository Dockerfile wraps the duck binary at /usr/bin/duck.

Is Cyberduck trustworthy?

The source is published under GPL-3.0 and the repository carries a SECURITY.md file for reporting. The README also warns that snapshot and beta builds from the development trunk are potentially unstable and experimental, and notes that the update channel is a user setting under Preferences, Update.

How do I use Cyberduck for FTP?

FTP is one of the protocols named in the repository description, and the desktop client for macOS and Windows is where connections are managed through bookmarks. The README directs readers to docs.cyberduck.io for general help.

Official sources

  1. Issues
  2. iterate-ch/cyberduck on GitHub
  3. License: GPL-3.0
  4. Project website
  5. README
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/iterate-ch-cyberduck.svg)](https://hysenlabs.com/projects/iterate-ch-cyberduck)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/iterate-ch-cyberduck.svg)](https://hysenlabs.com/projects/iterate-ch-cyberduck)

Was this useful?