Installation

Get started with AgentML by installing the agentmlx runtime. AgentML is a language specification—you write .aml files that define your agent's behavior, and agentmlx executes them.

Prerequisites

Linux or macOS (Windows support planned)
Architecture: amd64 or arm64
Internet connection for installation

Quick Install (Recommended)

Install the latest version with a single command:

curl -fsSL sh.agentml.dev | sh

This will:

Automatically detect your platform (Linux/macOS, amd64/arm64)
Download the latest release
Verify checksums for security
Install to ~/.agentmlx/bin
Add to your PATH

Installation Channels

Install from different release channels:

# Latest stable release (default)
curl -fsSL sh.agentml.dev | sh
 
# Next (release candidate)
curl -fsSL sh.agentml.dev | sh -s -- --channel next
 
# Beta releases
curl -fsSL sh.agentml.dev | sh -s -- --channel beta

Release Channels:

latest - Stable releases (v1.0.0) - Default
next - Release candidates (v1.0.0-rc.1)
beta - Beta releases (v1.0.0-beta.1)

The installer automatically falls back if a channel is empty: latest → next → beta

Install Specific Version

Install a specific version by providing the --version flag:

curl -fsSL sh.agentml.dev | sh -s -- --version 1.0.0-rc.1

Custom Installation Directory

By default, agentmlx installs to ~/.agentmlx/bin. To install elsewhere:

export AGENTMLX_INSTALL_DIR=/usr/local
curl -fsSL sh.agentml.dev | sh

Verify Installation

After installation, verify that agentmlx is working:

agentmlx --version

You should see output like:

agentmlx version 1.0.0-rc.1

Manual Installation

If you prefer to install manually, download the appropriate binary for your platform from GitHub Releases:

Linux (amd64)

curl -L https://amlx.agentml.dev/agentmlx/v1.0.0/agentmlx_1.0.0_linux_amd64 -o agentmlx
chmod +x agentmlx
sudo mv agentmlx /usr/local/bin/

macOS (amd64)

curl -L https://amlx.agentml.dev/agentmlx/v1.0.0/agentmlx_1.0.0_darwin_amd64 -o agentmlx
chmod +x agentmlx
sudo mv agentmlx /usr/local/bin/

macOS (arm64 / Apple Silicon)

curl -L https://amlx.agentml.dev/agentmlx/v1.0.0/agentmlx_1.0.0_darwin_arm64 -o agentmlx
chmod +x agentmlx
sudo mv agentmlx /usr/local/bin/

Environment Setup

API Keys

If you're using LLM namespaces like Gemini or Ollama, set up your API keys:

# For Gemini
export GEMINI_API_KEY=your_key_here
 
# For OpenAI (if using custom namespaces)
export OPENAI_API_KEY=your_key_here
 
# For Anthropic (if using custom namespaces)
export ANTHROPIC_API_KEY=your_key_here

Add these to your shell profile (~/.bashrc, ~/.zshrc, etc.) to persist them:

echo 'export GEMINI_API_KEY=your_key_here' >> ~/.zshrc

Project Structure

We recommend the following project structure for AgentML projects:

my-agent-project/
├── agents/              # Your AgentML files
│   ├── main.aml
│   ├── customer-support.aml
│   └── workflows/
│       └── onboarding.aml
├── schemas/             # JSON Schema definitions
│   └── events.json
├── scripts/             # External scripts for <script src="">
│   └── utils.js
├── tests/               # Test files
└── .env                 # Environment variables

Updating agentmlx

To update to the latest version, simply run the installer again:

curl -fsSL sh.agentml.dev | sh

Or update to a specific version:

curl -fsSL sh.agentml.dev | sh -s -- --version 1.1.0

Uninstalling

To uninstall agentmlx, remove the installation directory:

rm -rf ~/.agentmlx

Then remove the PATH entry from your shell profile:

# Edit ~/.bashrc, ~/.zshrc, or your shell profile
# Remove the line: export PATH="$PATH:$HOME/.agentmlx/bin"

Troubleshooting

Command not found

If you get command not found after installation:

Reload your shell configuration:

source ~/.bashrc  # or ~/.zshrc

Or manually add to PATH:

export PATH="$PATH:$HOME/.agentmlx/bin"

Permission denied

If you encounter permission errors during installation:

# Ensure the installation directory is writable
mkdir -p ~/.agentmlx/bin
chmod u+w ~/.agentmlx/bin

Checksum verification failed

If checksum verification fails:

Check your internet connection
Try downloading again
Report the issue if it persists: GitHub Issues

Apple Silicon / Rosetta 2

The installer automatically detects if you're running on Apple Silicon via Rosetta 2 and downloads the appropriate arm64 binary. If you encounter issues, manually specify:

AGENTMLX_PLATFORM=darwin_arm64 curl -fsSL sh.agentml.dev | sh

Build from Source

For development or contribution, you can build agentmlx from source:

git clone https://github.com/agentflare-ai/agentmlx.git
cd agentmlx
go build -o agentmlx main.go