// the take
A Terraform module that wraps nearly every aws_s3_bucket_* resource into one interface: versioning, SSE, lifecycle rules, CRR, access/CloudTrail/ELB/WAF log delivery, website hosting, object lock, plus newer surface like directory buckets, table buckets (Iceberg), S3 Vectors and S3 Files. Good fit for teams that want one module to stand up any flavor of S3 bucket instead of hand-writing each resource and its IAM policy JSON.
Pre-built IAM policy documents for the boilerplate everyone eventually writes by hand — deny insecure transport, deny unencrypted/SSE-C uploads, ELB/NLB/WAF/CloudTrail log delivery — so you flip a bool instead of authoring policy JSON. It tracks AWS's expanding S3 feature set rather than staying a basic bucket wrapper: directory buckets, table buckets, S3 Vectors, and a full S3 Files (EFS-like) submodule with its own IAM role, security groups and access points. Secure-by-default posture (block_public_acls/block_public_policy default true). The create_bucket flag is the standard, well-understood workaround for Terraform's lack of conditional module instantiation.
prevent_destroy and ignore_changes can't be passed in as module variables — that's a Terraform language limit (lifecycle blocks must be literals), not a design choice, but it means this module can't protect a bucket from an accidental destroy or stop an external tool's tags from being clobbered on every apply; you fork the module or work around it with an IAM deny statement instead. The README's fix for that gap points to the maintainer's own paid product (compliance.tf) rather than shipping the workaround in the open module, which is a conflict of interest worth knowing about before you lean on this module's docs for that problem. The input surface is enormous — deeply nested optional objects for file_systems, analytics_configuration, intelligent_tiering, inventory_configuration, etc. — so diffing a version bump or onboarding a new feature means reading a lot of variable schema. S3 Files has sharp edges buried in prose, not enforced by the schema: changing a file system's prefix, KMS key, or IAM role replaces it along with its mount targets and access points, access points can't be edited at all (only replaced), and dropping a synchronization_configuration silently changes data expiry from 30 to 365 days.
We read the README, the directory tree and a key source file in November 2026. Nobody pays to be listed. How we judge →