Show HN: Gat – Version large files with Git, without an LFS server

Gat is a tool for versioning large files alongside Git without needing a special LFS server. It uses object stores (S3, Azure Blob, GCS) or plain directories as backends: gat add records large files into a gat.lock and shard/cache layout, and gat sync reconciles the working tree with that lock instead of relying on Git clean/smudge filters. Git hooks installed by gat init run sync after checkout, merge/pull, and rebase/amend so commit, push, pull and clone workflows keep working normally; each clone must run gat init and use gat pull to download content missing from its cache.

Installation is available via one-line shell or PowerShell installers or by building from source with Cargo (requires Rust ≥ 1.91). Core commands include gat remote add to configure storage URLs, gat push/pull for transfers, and gat sync for local reconciliation; remotes are set per transfer unless a default is chosen. The project is open-source under Apache-2.0, documented at getgat.dev, and the repository contains multiple crates, tests, benches and example usage demonstrating how to track, push and pull large assets while keeping Git workflows unchanged.