Really, really fast builds.
The quickest way to get started with Hurry is to use our managed build cache. To start, you'll need to sign up for an account at hurry.build to get your HURRY_API_TOKEN. Then, run:
# Instead of `cargo build`:
$ HURRY_API_TOKEN=your_token_here hurry cargo build
Alternatively, you can self-host Hurry locally or on your own infrastructure.
curl -sSfL https://hurry.build/install.sh | bash
You can configure certain installer options:
# Install to a specific directory (defaults to ~/.local/bin)
curl -sSfL https://hurry.build/install.sh | bash -s -- -b /usr/local/bin
# Install a specific version
curl -sSfL https://hurry.build/install.sh | bash -s -- -v 0.2.0
# Get help
curl -sSfL https://hurry.build/install.sh | bash -s -- -h
irm https://hurry.build/install.ps1 | iex
You can configure certain installer options:
# Install a specific version
$env:Version="0.2.0"; irm https://hurry.build/install.ps1 | iex
# Install to a custom directory (defaults to $env:LOCALAPPDATA\Programs\hurry)
$env:BinDir="C:\Tools"; irm https://hurry.build/install.ps1 | iex
# Show help
$env:Help="true"; irm https://hurry.build/install.ps1 | iex
Download the latest release for your platform from GitHub Releases, extract the archive, and place the hurry binary in your PATH.
Hurry works by examining the build plan generated by Cargo, seeing if any of the necessary artifacts are restorable from remote cache, and downloading them into your target folder if they're available. It then runs the build and uploads any missing artifacts to the remote cache.
Hurry is currently in alpha, and there are many rough edges. Some current known limitations include:
cargo commands other than cargo build (e.g. cargo run, cargo test, cargo install).We are currently working on resolving all of these limitations. If any of them are a blocker for your team's adoption of Hurry, please reach out to us and we can help prioritize your rollout.
Like Cargo, Hurry supports a wide range of platforms using a target tier policy. The supported platforms are:
| Platform | Architecture | Target Triple |
|---|---|---|
| macOS | x86_64 (Intel) | x86_64-apple-darwin |
| macOS | ARM64 (Apple Silicon) | aarch64-apple-darwin |
| Linux | x86_64 (glibc) | x86_64-unknown-linux-gnu |
| Linux | ARM64 (glibc) | aarch64-unknown-linux-gnu |
| Linux | x86_64 (musl) | x86_64-unknown-linux-musl |
| Linux | ARM64 (musl) | aarch64-unknown-linux-musl |
| Windows | x86_64 | x86_64-pc-windows-gnu |
Hurry provides Tier 1 support for 2 platforms:
aarch64-apple-darwinx86_64-unknown-linux-gnuThese platforms have full support for build acceleration and distributed caching.
Hurry provides Tier 2 support for 3 platforms:
aarch64-unknown-linux-gnux86_64-unknown-linux-muslaarch64-unknown-linux-muslPlatforms in Tier 2 have unstable support for build acceleration and distributed caching. We intend to support build acceleration on these platforms, and we have automate testing coverage over them, but we do not exercise their hot paths as extensively as Tier 1 platforms, so their support may be less stable.
Hurry provides Tier 3 support for 1 platform:
x86_64-pc-windows-gnuPlatforms in Tier 3 have automated testing and precompiled binaries, but their subcommands may not be accelerated. For example, hurry cargo build on Windows will directly pass through to cargo build, and does not implement any distributed caching.
The intent of Tier 3 platforms is to help your team migrate your build processes to hurry more easily. Even in platforms where hurry does not yet support build acceleration, you can still replace all instances of cargo build with hurry cargo build, and your build should still behave the same as before.
Hurry is licensed under the Apache 2.0 License.
$ claude mcp add hurry \
-- python -m otcore.mcp_server <graph>