Installation

Stella ships as a single self-contained binary named stella. There is no runtime to install alongside it.

Prerequisites

• macOS or Linux, x86_64 or arm64.
• For the prebuilt / Homebrew paths: nothing but curl.
• For cargo install / building from source: Rust 1.90+ (via rustup) and git.
• An API key for any one supported provider — or a local OpenAI-compatible model server (Ollama, vLLM, LM Studio, llama.cpp) and no key at all. See Providers & models.
• Optional: ripgrep and fd on your PATH (the grep/glob tools shell out to them), and gh for the CI and issue tools.

Install

Prebuilt binary (curl | sh)

Installs the latest tagged release for your platform, verifies its SHA-256, and falls back to cargo install on platforms without a prebuilt binary (unsupported targets, and musl-libc Linux even on supported architectures):

curl -fsSL https://raw.githubusercontent.com/macanderson/stella/main/install.sh | sh
stella --version

The binary lands in ~/.local/bin by default — if stella isn't found afterwards, add that directory to your PATH. Two env vars tune the script: STELLA_INSTALL_DIR overrides the destination, STELLA_VERSION pins a specific release tag:

STELLA_INSTALL_DIR=/usr/local/bin STELLA_VERSION=v0.4.35 \
  sh -c "$(curl -fsSL https://raw.githubusercontent.com/macanderson/stella/main/install.sh)"

Homebrew

Installs the prebuilt binary from the tap — no Rust toolchain needed. The release workflow keeps the formula in sync on every tag:

brew install macanderson/tap/stella
# equivalently:
brew tap macanderson/tap && brew install stella

From cargo

Requires Rust 1.90+ and git:

cargo install --locked --git https://github.com/macanderson/stella stella-cli
stella --version

From source

git clone https://github.com/macanderson/stella.git
cd stella
cargo build --release
./target/release/stella --version

Verify

stella version

Then confirm a provider is configured:

stella models   # every provider, its default model, key status, and base URL

Next: the Quickstart.