s7cmd

A reliable, flexible, and fast command-line tool for Amazon S3.

s7cmd combines the speed of a Rust async runtime with the breadth of the AWS S3 API surface, providing high-throughput object operations (ls, cp, mv, rm, sync, clean) alongside comprehensive bucket administration (lifecycle, policy, encryption, CORS, public access block, website, logging, notification, and more) — all from a single static binary.

s7cmd is a thin command-line wrapper over four Rust libraries by the same author: s3sync, s3ls-rs, s3util-rs, and s3rm-rs. Originally, only s3sync and s3rm-rs existed and were not intended to be merged, but in response to user requests for a unified interface, the functionality was split into focused libraries and bundled together into a single binary as s7cmd. Built on aws-sdk-rust and tokio, s7cmd targets workloads that demand both performance and operational completeness: data engineering pipelines, ML training data preparation, multi-account bucket governance, and integrity-critical migrations.

Why s7cmd?

• Single binary, full coverage. Object transfer, bulk delete, and every common bucket-level configuration in one tool.
• Strong integrity verification. Native support for SHA256, SHA1, CRC32, CRC32C, and CRC64NVME — aligned with S3's 2025 default checksum policy.
• Predictable performance. Configurable workers, multipart thresholds, and chunk sizes; bounded memory footprint suitable for small instances and large CI runners alike.
• Apache-2.0 licensed. No copyleft concerns for enterprise deployment or container distribution.

Scope

s7cmd is designed to cover Amazon S3 object operations and bucket management — listing (ls), single- and bulk-object transfers (cp / mv / rm), recursive synchronization (sync), bulk delete (clean), and the common bucket-level configurations (tagging, versioning, policy, lifecycle, encryption, CORS, public-access-block, website, logging, notification). For any S3 use case outside that scope, use a more comprehensive tool such as the AWS CLI (aws s3 / aws s3api).

s7cmd targets Amazon S3 as its primary supported platform. S3-compatible storage (MinIO, Cloudflare R2, Backblaze B2, Wasabi, Ceph RGW, DigitalOcean Spaces, IBM COS, and similar) is supported on a best-effort basis only — such services may work via --endpoint-url, but they are not part of the official test matrix and behavior may change between releases. This is a structural consequence of building on aws-sdk-rust, which is generated from AWS service models and assumes Amazon S3 semantics (checksum headers, endpoint resolution, signing variants, response schemas); features that depend on AWS-specific semantics, such as CRC64NVME checksums or newer S3 API additions, may not work against non-AWS endpoints. Bug reports against S3-compatible storage will be triaged but not prioritized, and fixes are not guaranteed.

s7cmd is not intended to be a drop-in replacement for, or behaviorally compatible with, any other S3 client — including the AWS CLI (aws s3, aws s3api) and tools such as s3cmd, s4cmd, s5cmd, and s6cmd. Its command-line flags, transfer semantics, verification rules, and exit codes are designed around the underlying libraries' own scope and design principles — not interoperability with another tool's interface. Output formats and flag names will not be adjusted to match any external tool, and scripts written against another S3 client should not be expected to work with s7cmd unmodified. The numeric progression in the name (s3cmd → s4cmd → s5cmd → s6cmd → s7cmd) does not imply succession or compatibility.

Non-Goals

The following are explicitly out of scope and will not be added, regardless of demand:

• Official support, testing, or guaranteed compatibility for any storage service other than Amazon S3. S3-compatible storage may work on a best-effort basis as described in the Scope section above, but adding dedicated code paths, provider-specific workarounds, or backends for services such as MinIO, Cloudflare R2, Backblaze B2, Wasabi, Ceph RGW, DigitalOcean Spaces, IBM COS, Tencent COS, Alibaba OSS, Azure Blob Storage, or Google Cloud Storage is out of scope.
• Feature parity with, or porting features from, other S3 clients. Feature requests of the form "tool X has feature Y, please add it to s7cmd" — including variants such as "feature Y would also be useful in s7cmd," "many users expect Y because tool X has it," or "Y is missing compared to tool X" — will be closed without further discussion. The existence of a feature, flag, command, output format, or behavior in aws s3, aws s3api, s3cmd, s4cmd, s5cmd, s6cmd, or any other S3 tool carries no weight in s7cmd's design decisions, regardless of how the request is framed. Each feature is evaluated solely against s7cmd's own scope and the design principles of its underlying libraries. If the feature you need exists in another tool, use that tool.
• FUSE filesystem mounting, daemon mode, or any persistent background process. s7cmd is a one-shot CLI; it runs, transfers, and exits.
• Workflow orchestration features — scheduling, cross-run state databases, retry queues that survive process restart, or DAG execution. Use a workflow engine such as Airflow, Argo Workflows, or AWS Step Functions for orchestration.
• A graphical user interface, a TUI, or an interactive shell mode.
• A plugin or extension mechanism.
• AWS service coverage beyond S3. s7cmd will not add subcommands for IAM, KMS, CloudFront, or any other AWS service, even when they interact closely with S3.

Issues and pull requests requesting any of the above will be closed.

Maintenance Model

s7cmd is maintained as a personal project. Dependency updates and critical bug fixes are applied on a best-effort basis. New features are not actively solicited. If you need guaranteed enterprise support, this is not the tool for you.

About the name

The name follows the s3cmd / s4cmd / s5cmd / s6cmd lineage, but s7cmd is not affiliated with, derived from, or compatible with any of them. The number 7 was chosen simply because it was the next available one. There is no deeper meaning.

Usage

Usage: s7cmd [OPTIONS] [COMMAND]

Object Operations:
  ls                                    List S3 objects
  cp                                    Copy objects from/to S3 (or S3 to S3)
  mv                                    Move objects from/to S3 (copy then delete source)
  rm                                    Delete a single S3 object
  sync                                  Synchronize files between local and S3 (or S3 to S3)
  clean                                 Bulk-delete S3 objects

Object Metadata:
  head-object                           Head an S3 object
  get-object-tagging                    Get an S3 object's tagging
  put-object-tagging                    Put tagging on an S3 object
  delete-object-tagging                 Delete tagging from an S3 object

Bucket Operations:
  create-bucket                         Create an S3 bucket
  delete-bucket                         Delete an S3 bucket
  head-bucket                           Head an S3 bucket

Bucket Tagging:
  get-bucket-tagging                    Get a bucket's tagging
  put-bucket-tagging                    Put tagging on a bucket
  delete-bucket-tagging                 Delete tagging from a bucket

Bucket Policy:
  get-bucket-policy                     Get a bucket's policy
  put-bucket-policy                     Put a bucket policy
  delete-bucket-policy                  Delete a bucket's policy

Bucket Versioning:
  get-bucket-versioning                 Get a bucket's versioning configuration
  put-bucket-versioning                 Put a bucket versioning configuration

Bucket Lifecycle:
  get-bucket-lifecycle-configuration    Get a bucket's lifecycle configuration
  put-bucket-lifecycle-configuration    Put a bucket lifecycle configuration
  delete-bucket-lifecycle-configuration Delete a bucket's lifecycle configuration

Bucket Encryption:
  get-bucket-encryption                 Get a bucket's encryption configuration
  put-bucket-encryption                 Put a bucket encryption configuration
  delete-bucket-encryption              Delete a bucket's encryption configuration

Bucket CORS:
  get-bucket-cors                       Get a bucket's CORS configuration
  put-bucket-cors                       Put a bucket CORS configuration
  delete-bucket-cors                    Delete a bucket's CORS configuration

Bucket Public Access Block:
  get-public-access-block               Get a bucket's public access block configuration
  put-public-access-block               Put a bucket public access block configuration
  delete-public-access-block            Delete a bucket's public access block configuration

Bucket Website:
  get-bucket-website                    Get a bucket's website configuration
  put-bucket-website                    Put a bucket website configuration
  delete-bucket-website                 Delete a bucket's website configuration

Bucket Logging:
  get-bucket-logging                    Get a bucket's logging configuration
  put-bucket-logging                    Put a bucket logging configuration

Bucket Notification:
  get-bucket-notification-configuration Get a bucket's notification configuration
  put-bucket-notification-configuration Put a bucket notification configuration

Other:
  help                                  Print this message or the help of the given subcommand(s)

Options:
      --auto-complete-shell <SHELL>  Generate shell completions for s7cmd (all subcommands) and exit [possible values: bash, elvish, fish, powershell, zsh]
  -h, --help                         Print help (see more with '--help')
  -V, --version                      Print version

Documentation

Each subcommand is documented in the README of its underlying library. For details on flags, semantics, and exit codes, refer to:

Each of these projects also ships its own standalone binary, which can be used independently of s7cmd.

Requirements

• x86_64 Linux (kernel 3.2 or later)
• ARM64 Linux (kernel 4.1 or later)
• Windows 11 (x86_64, aarch64)
• macOS 11.0 or later (aarch64, x86_64)

All features are tested on the above platforms.

Installation

Download the latest binary from GitHub Releases

You should build Intel Mac and ARM64 Windows binaries yourself.

Fully AI-generated, always human-verified

No human wrote a single line of source code in this project. Every line of s7cmd's own source code (including the vendored adaptations from upstream), every test, all documentation, CI/CD configuration, and this README were generated by AI using Claude Code (Anthropic). The same applies to three of the four underlying libraries: s3util-rs, s3ls-rs, and s3rm-rs. The fourth, s3sync, is human-written and serves as the reference architecture from which the AI-generated siblings were derived.

Human verification is a permanent policy, not a one-time event applied only to the initial build. Human engineers authored the requirements, design specifications, and s3sync reference architecture, and continue to review and verify every change to the design, source code, and tests. Every release is manually tested by humans before it ships, and all E2E test scenarios are verified against live AWS S3. No AI-generated change is released without human review and testing — this applies equally to the initial build and to all future updates, including dependency bumps, bug fixes, and new features. The development follows a spec-driven process: requirements and design documents are written first, and the AI generates code to match those specifications under continuous human oversight.

Every underlying library maintains 96%+ automated test coverage. This serves a dual purpose: it verifies that AI-generated code meets its specifications, and it ensures the project remains maintainable by hand — whether because AI tooling becomes unavailable, or because a future maintainer prefers to work without AI assistance. Combined with the modular library design and Apache-2.0 licensing, this means s7cmd can be safely forked and maintained without AI assistance if the need arises.

License

Apache-2.0. See LICENSE.