S3 Browser FAQ

Brows3 is a free, open-source desktop browser for Amazon S3 and S3-compatible object storage. It provides bucket and prefix navigation, search, sorting, previews, editing, transfers, object properties, and ACL controls through a Rust and Tauri application.

Yes. Brows3 is released under the MIT license with no subscription, trial period, account signup, or paid feature tier. The source and release artifacts are public on GitHub.

Brows3 publishes Windows x64 MSI, EXE, and portable ZIP packages, macOS DMGs for Apple Silicon and Intel, and Linux AppImage and Debian packages for Intel/AMD 64-bit and ARM64. Linux filenames marked amd64 work on both Intel and AMD 64-bit systems. An experimental i386 Debian package targets i686 Linux systems; it requires WebKitGTK 4.1 and manual updates, and real-device testing remains limited. There is no 32-bit Windows installer or Cargo prebuilt binary. Check Downloads for the packages in the current release. Linux requires GTK 3.24 and WebKitGTK 4.1.

Brows3 supports Amazon S3 plus S3-compatible services such as MinIO, Cloudflare R2, Wasabi, DigitalOcean Spaces, Garage, Backblaze B2, STACKIT Object Storage, and custom endpoints. Provider-specific behavior can vary when an implementation does not support an S3 API feature such as ACLs.

Use manual access-key credentials, select a shared AWS profile from your system configuration, or choose an AWS IAM Identity Center profile. Restricted users can open a known s3:// bucket path without requiring ListBuckets permission.

Yes. Brows3 identifies current sso_session profiles and legacy SSO profiles, then resolves cached temporary credentials through the AWS SDK. If the session is missing or expired, the profile dialog can run AWS CLI v2 browser sign-in for that exact profile and retest it.

Brows3 uses macOS Keychain, Windows Credential Manager, or freedesktop Secret Service on Linux. Linux also migrates a still-live credential from the old volatile keyutils backend. Portable mode and native-keychain failures use a local plaintext secrets.json fallback, restricted to mode 0600 on Unix.

Yes. Files at or above 100 MiB use multipart upload with adaptive part sizes, bounded memory, retryable parts, per-part progress, cancellation, and remote abort cleanup. This supports files beyond S3's 5 GiB single-PUT limit while staying within the 10,000-part limit.

Current-folder filtering works on the loaded listing. Deep search lists the selected S3 prefix recursively and is deliberately bounded to 100 LIST requests, 100,000 scanned objects, and 10,000 returned matches per search. It does not rely on a hidden full-bucket background index.

Bucket discovery is cached per profile for 30 minutes. Non-default complete-prefix sort results use a bounded in-memory session cache that is invalidated after writes or manual refreshes. Normal folder browsing remains prefix-aware and paginated. Brows3 does not build a complete bucket tree in the background.

Yes. Use All Buckets to create a bucket, edit its JSON policy or delete an empty bucket. These actions require the corresponding provider permissions. Deletion requires the exact bucket name, and switching profiles closes pending bucket actions.

Yes. Add a bucket from its header or the sidebar. Bucket and folder favorites belong to the selected profile and reopen the saved location.

The persisted text, HTML, and code preview limit can be set from 1 to 100 MB, with a 2 MB default. The Rust backend streams and enforces that exact bound even when S3 metadata is missing. Images, audio, video, and PDFs use presigned streaming and do not consume the text-preview allowance.

Yes. Uploads infer a MIME type from the object name, and Properties provides a validated free-form Content-Type editor with common suggestions. Metadata-only changes preserve supported tags, user metadata, headers, encryption and object-lock fields, and ACL grants. Objects above 5 GiB use multipart copy.

Yes. When a shared AWS profile resolves an endpoint_url, Brows3 enables path-style bucket addressing and compatibility checksum behavior, matching profiles created through Custom S3 mode.

Run winget install --exact --id Brows3Team.Brows3. The public catalog can briefly lag behind the latest GitHub release while Microsoft's validation and indexing finish.

Run cargo install brows3 --locked, then launch brows3 from your terminal. The crates.io package contains the compiled web interface, so Node.js and pnpm are not required. Building still needs Rust and the native platform toolchain and libraries. If cargo-binstall is already installed, cargo binstall brows3 downloads a prebuilt executable for supported macOS, Linux and Windows 64-bit targets. Linux needs GTK 3, WebKitGTK 4.1 and a graphical desktop; Windows needs WebView2. Cargo installs an executable rather than a desktop shortcut or macOS app bundle.

Cargo builds and the experimental 32-bit Linux package do not use the built-in desktop updater. For Cargo, rerun cargo install brows3 --locked or cargo binstall brows3 to install a newer published version. For 32-bit Linux, download the newer i386 Debian package and install it through your system package manager. Use the same installation method to avoid confusing multiple copies of the application.

An S3 browser is a graphical client for inspecting buckets and managing objects through an S3 API. Brows3 runs as a desktop application, not a hosted file manager in your web browser. Use it to inspect a prefix, preview an object, edit text, and upload or download files. S3 folders are key prefixes, not ordinary filesystem directories. Brows3 is the name of this open-source project; it is not affiliated with other products named S3 Browser.

Install the build for your operating system, create a connection profile, and select the AWS connection method that matches your credentials. Use an existing shared AWS profile or access credentials with the correct region; SSO profiles use AWS CLI v2 sign-in when needed. Test the connection, open a bucket, and start with a narrow prefix. If your credentials cannot list all buckets, open a known s3://bucket/prefix path instead. Never use account-root credentials just to make a connection test pass, and begin with read-only access when you only need to inspect objects.

Yes. Configure the provider or Custom S3 connection with the service's S3 API endpoint, region where required, access key ID and secret access key. For MinIO, use the S3 API endpoint rather than the separate web-console URL. For Cloudflare R2, use the S3 endpoint and S3 API credentials from your account, not a public bucket URL or a general Cloudflare API token. Custom S3 connections use path-style addressing and compatibility checksum behavior. Provider support for ACLs and bucket policies differs, so an unsupported administration action does not necessarily mean object browsing is broken.

Check the selected profile, region, endpoint and credential expiry first. A successful sign-in does not grant access to every bucket. Bucket listing and object access are separate permissions; try a known bucket path if account-wide listing is restricted. For AWS SSO, refresh the session through AWS CLI v2 and retest the same profile. For encrypted objects, your identity may also need permission to use the encryption key. Do not make a bucket public as a troubleshooting shortcut. Ask the storage administrator to review access to the exact bucket, prefix and action that failed.

Choose Brows3 for interactive browsing, visual previews, one-off text edits and transfers across S3-compatible services. Use the AWS Console for AWS-specific administration that Brows3 does not expose. Use the AWS CLI or another automation tool for scripts, scheduled jobs and reproducible bulk workflows. A desktop client is not a replacement for a backup schedule, retention policy or restore test. Start transfers on a small test prefix before changing production data.

Brows3 is free under the MIT license, but your storage provider can charge for requests, storage and data transfer. Opening folders issues listing requests; downloading or streaming previews can transfer object data. Deep search is bounded rather than a permanent whole-bucket index, but still issues requests. Check your provider's pricing and narrow the prefix when searching large buckets. Brows3 does not remove provider-side costs or limits.

macOS may require manual approval for apps downloaded outside the App Store. Check the release notes for signing and notarization details for the version you download.

The Downloads section links directly to the installers and portable packages attached to the latest stable GitHub release, grouped by system and architecture. Updater archives and signatures remain available on the GitHub release page.