blitcp

blitcp is a command-line file copier — with an optional desktop GUI — built around one idea: a copy that reports success should have actually succeeded. Every copied file is read back and hashed against the source, and the tool returns exit codes a script can act on, so a drive that writes the wrong bytes is caught instead of silently trusted.

It is built for people who move large amounts of data: backups to USB drives and NAS boxes, server-to-server migrations, and scheduled cron jobs. It reads files in physical disk order so a spinning disk stops seeking, and it deduplicates by content hash — identical files are copied once and the rest become hard links or reflinks, so a tree full of duplicates costs the space of one copy. For remote transfers it streams tar over a raw SSH channel instead of paying SFTP's per-file round trip, and sparse files are copied through SEEK_DATA/SEEK_HOLE so holes never reach the wire.

Local disks, SSH/SFTP servers, SMB/CIFS shares, and cloud object storage (Amazon S3 and any S3-compatible store — Cloudflare R2, Backblaze B2, Wasabi, MinIO and others via built-in presets — plus Azure Blob and Google Cloud Storage) are all ordinary endpoints, usable in any combination. A --sftp-only mode handles locked-down gateways that close the exec channel.

It is made to live in scripts: a --quiet mode prints one line on success for cron and pipelines, --progress keeps the bar while staying quiet otherwise, and there is no telemetry and no network call beyond the connection you explicitly ask for. Both the CLI and the desktop GUI are fully translated into six languages besides English — German, Greek, Spanish, Italian, Japanese and Simplified Chinese. Free and open source under Apache-2.0, with no paid tier.

Installation: pip install blitcp (PyPI), piwheels for Raspberry Pi, the AUR (blitcp-bin / blitcp-gui-bin) for Arch, and standalone release binaries for Linux, macOS and Windows that need no Python installed.