What is this?
gat is a simple, fast, versioned large-file storage tool for git: point it at S3, Azure Blob, GCS, or a plain directory, and your normal git add/commit/push/pull/clone commands keep working as before for large files.
gat is a simple, fast, versioned large-file storage tool for git: point it at S3, Azure Blob, GCS, or a plain directory, and your normal git add/commit/push/pull/clone commands keep working as before for large files.Is this a standalone CLI, or does it need a special server like git-lfs?It's a plain command-line tool with no required special server — it installs git hooks that keep your working tree in sync with a gat.lock file instead of using clean/smudge filters.
It's a plain command-line tool with no required special server — it installs git hooks that keep your working tree in sync with a gat.lock file instead of using clean/smudge filters.How do I get it running in my repo?Install it with the Linux/macOS shell script or the Windows PowerShell command (or `cargo install` from source), then run `gat init` inside your git repo, `gat add` the large files, and point a remote at your storage bucket.
Install it with the Linux/macOS shell script or the Windows PowerShell command (or `cargo install` from source), then run `gat init` inside your git repo, `gat add` the large files, and point a remote at your storage bucket.Do I need Rust or special credentials to use it?Prebuilt installs just need the install script and your storage credentials; building from source requires Rust 1.91 or newer, which is the project's tracked minimum supported Rust version.
Prebuilt installs just need the install script and your storage credentials; building from source requires Rust 1.91 or newer, which is the project's tracked minimum supported Rust version.If I add a remote, does gat automatically use it?No — adding a remote doesn't make it the default, so you have to pass --remote origin on every transfer or explicitly run `gat remote default origin`.
No — adding a remote doesn't make it the default, so you have to pass --remote origin on every transfer or explicitly run `gat remote default origin`.How does gat sync files without git's clean/smudge filters?gat sync reconciles your working tree directly with the current gat.lock file or shard directory, and git hooks installed by `gat init` run sync automatically after checkout, merge/pull, and rebase/amend.
gat sync reconciles your working tree directly with the current gat.lock file or shard directory, and git hooks installed by `gat init` run sync automatically after checkout, merge/pull, and rebase/amend.What is the Gat MCP server for?It connects your AI coding tool directly to gat's documentation so it can look up usage and command reference while you work.
It connects your AI coding tool directly to gat's documentation so it can look up usage and command reference while you work.How is gat different from git-lfs or dvc?Unlike git-lfs, gat doesn't require a special server — any object store or plain directory works; unlike dvc, it focuses on one job, keeping large-file storage transparent to your normal git workflow.
Unlike git-lfs, gat doesn't require a special server — any object store or plain directory works; unlike dvc, it focuses on one job, keeping large-file storage transparent to your normal git workflow.Which topics is it in?google-cloud, object-storage, cloud-storage, azure, s3, version-control, storage, cli-tool, git, command-line, devops, mcp-server
google-cloud, object-storage, cloud-storage, azure, s3, version-control, storage, cli-tool, git, command-line, devops, mcp-server