Install

Run it on your hardware.

Use a prebuilt release on GNU/Linux with glibc 2.35 or newer, or on macOS 11 or newer. Windows runs through WSL2. Axocoatl needs Podman for the default local Session sandbox.

Recommended

Install the latest release.

The installer detects GNU/Linux or macOS on x86_64 or ARM64, downloads the matching GitHub release, and requires and verifies its published SHA-256 file. Before download it requires Linux to report glibc 2.35 or newer and macOS to report version 11 or newer.

curl -fsSL https://axocoatl.ai/install.sh | sh

Review the installer source before piping it to a shell.

Requirements
  • GNU/Linux with glibc 2.35+, or macOS 11+
  • Windows through WSL2
  • Podman for local Session sandboxes
  • An Ollama or hosted model provider
Other paths

Build it yourself.

Cargo

Install from crates.io

Requires Rust 1.88 or newer.

cargo install axocoatl-cli

Source

Build this checkout

Build the explicit CLI package. The root package is only a benchmark placeholder.

cargo build -p axocoatl-cli --release

Windows

Use WSL2

Run wsl --install in an administrator PowerShell, then install Podman and Axocoatl inside the Linux distro.

first run
axocoatl onboard  # choose a provider
axocoatl doctor   # inspect the environment
axocoatl dev      # start daemon + app

open http://localhost:8080
First run

Verify before you work.

onboard creates a starting configuration. doctor validates configuration and data access, probes configured Ollama, checks credential presence for OpenAI, OpenRouter, Anthropic, Gemini, and Mistral, checks the Podman CLI and applicable machine state, requires a bounded podman info --format json response before reporting Podman ready, and checks daemon IPC. dev starts the daemon, IPC server, and browser app.

Open the local app, authorize a project directory, and create a Session from the workspace rail. Review its proposed runtime and setup command before anything in the repository runs.

Follow the complete quickstart →

Bring your model choices.

Configure Ollama, OpenAI, OpenRouter, Anthropic, Mistral, or Gemini. Each Agent can use the provider and model that fit its work.

Get started →OpenRouter setup