Skip to content

Install Axocoatl

Axocoatl publishes release binaries for Linux and macOS 11 or newer on x86_64 and aarch64. Windows runs through WSL2; there is no native Windows binary. The Linux archives target GNU libc 2.35 or newer. They do not run on musl-based distributions such as Alpine; use the Cargo or source path there.

Terminal window
curl -fsSL https://raw.githubusercontent.com/axocoatl/axocoatl/main/scripts/install.sh | sh

The installer resolves the latest GitHub release, downloads the tarball for the current OS and architecture, and requires its matching SHA-256 file. A missing, malformed, or mismatched checksum stops installation. The script also stops before download when Linux does not report GNU libc 2.35 or newer, or when macOS is older than version 11. It installs to /usr/local/bin when writable and otherwise to ~/.local/bin.

Verify that your shell can find the binary:

Terminal window
axocoatl --version
axocoatl --help

If installation used ~/.local/bin, add it to your shell path:

Terminal window
export PATH="$HOME/.local/bin:$PATH"

The binary can start without every optional dependency, but the normal local workbench path uses:

  • Podman for folder-anchored Session tools and parallel attempts;
  • at least one configured model provider;
  • Ollama plus a pulled model for the simplest local first run.

For a local Ollama setup:

Terminal window
ollama serve
ollama pull llama3.2

Run ollama serve in its own terminal or as its own service.

  1. In an administrator PowerShell, install WSL2 and reboot when requested:

    Terminal window
    wsl --install
  2. Open the Linux distribution and install Podman. On Ubuntu:

    Terminal window
    sudo apt-get update
    sudo apt-get install -y podman
  3. Run the release installer inside WSL2, then continue with axocoatl onboard.

axocoatl dev runs inside WSL2. Open http://localhost:8080 from the Windows browser; WSL2 normally forwards that loopback port to the host.

Create a configuration and run doctor →