Skip to main content

Getting Started

Get up and running with AgentVisor™ in minutes. The path has four steps:

  1. Get a license key: AgentVisor is a licensed product
  2. Install the prerequisites: Docker, Python, jq
  3. Install the CLI
  4. Activate your license on first run

Step 1: Get a License Key​

Start here: you need a license key

Released AgentVisor binaries and container images require a valid Manetu license key to run. Request a license (trial or purchase), or contact sales@manetu.com.

You don't need the key until Step 4. Installing the CLI and scaffolding a project with agentvisor template create work without one, so you can continue with Steps 2 and 3 while your request is processed. agentvisor serve, run, exec, and built images do require it.

Step 2: Install the Prerequisites​

Docker​

AgentVisor uses Docker to build, test, and deploy your agents.

# Verify Docker is installed and running
docker version

Install Docker Desktop:

Python 3.10+ (for LangGraph agents)​

Python is required for the built-in LangGraph framework provider and the tutorial. AgentVisor's framework provider architecture supports additional agent frameworks and languages.

# Check your Python version
python3 --version

Install Python:

# macOS
brew install python@3.12

# Ubuntu/Debian
sudo apt install python3.12 python3.12-venv

# Or download from https://www.python.org/downloads/

jq​

The tutorial and several example commands pipe API responses through jq to extract and pretty-print JSON fields.

# Verify jq is installed
jq --version

Install jq:

  • macOS: brew install jq
  • Ubuntu/Debian: sudo apt install jq
  • Other platforms: see the jq download page

Temporal and Ollama​

Projects created with agentvisor template create ship a compose.yml that brings up Temporal (durable workflow orchestration for agent runs) and Ollama (a local LLM server for the tutorial's default agent). You don't install these yourself — docker compose up -d starts them as containers alongside Docker.

Step 3: Install the CLI​

Install Script​

curl -fsSL https://raw.githubusercontent.com/manetu/agentvisor/main/install.sh | sh

The script detects your OS and architecture, downloads the correct binary, verifies its SHA256 checksum, and installs to /usr/local/bin (using sudo only if needed).

Environment variables:

VariableDefaultDescription
AGENTVISOR_VERSIONlatestSpecific version to install (e.g. 1.2.3, the v prefix is optional)
AGENTVISOR_INSTALL_DIR/usr/local/binInstallation directory
AGENTVISOR_NO_VERIFY0Set to 1 to skip checksum verification

Example with custom options:

AGENTVISOR_VERSION=1.2.3 AGENTVISOR_INSTALL_DIR=~/.local/bin \
curl -fsSL https://raw.githubusercontent.com/manetu/agentvisor/main/install.sh | sh

Homebrew (macOS / Linux)​

If you use Homebrew, tap the official AgentVisor formula and install:

brew tap manetu/homebrew-tap
brew install manetu/tap/agentvisor

Upgrade later with:

brew update
brew upgrade manetu/tap/agentvisor

Manual Download​

Download the tarball for your platform directly from GitHub Releases:

# Example for macOS ARM64 — replace VERSION with a tag from the Releases page
VERSION=v1.2.3
curl -sLO "https://github.com/manetu/agentvisor/releases/download/${VERSION}/agentvisor-darwin-arm64-${VERSION}.tar.gz"
tar xzf "agentvisor-darwin-arm64-${VERSION}.tar.gz"
sudo mv agentvisor /usr/local/bin/

Other platforms follow the same agentvisor-<os>-<arch>-<version>.tar.gz pattern (linux-amd64, linux-arm64, darwin-amd64). Each release also publishes a checksums.txt you can verify against, and a .whl for the Python SDK.

Verify Installation​

agentvisor --help

Expected output:

Manetu AgentVisor™ is a Controlled Autonomous Runtime Environment providing
confinement and observability for AI agents. It automatically detects your application
runtime and builds a combined image with the host runtime, guest proxy, and your agent code.

Usage:
agentvisor [command]

Available Commands:
build Build a combined AgentVisor™ image from your agent source
run Run a single agent execution and exit
serve Start AgentVisor™ and wait for API requests
exec Start an interactive terminal session inside an AgentVisor sandbox
schema Generate schema manifest for agent graphs
template Manage agent project templates
mcp Interact with configured MCP servers
a2a Interact with configured A2A agents
completion Generate the autocompletion script for the specified shell
help Help about any command

Flags:
--config stringArray Config overlay file(s) to merge (can be repeated; later files override earlier)
-h, --help help for agentvisor
-q, --quiet Suppress all output except errors
-v, --verbose Show detailed output including debug info
--version version for agentvisor

Use "agentvisor [command] --help" for more information about a command.

Step 4: Activate Your License​

The first time you run a licensed command (agentvisor serve, run, or exec) on a terminal, the CLI prompts for your key:

AgentVisor license key: ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
(Enter to submit, Ctrl+C to cancel)
Get a license: https://agentvisor.dev/licensing

Paste the key and press Enter. AgentVisor validates it online and stores it in your OS keyring, so you aren't prompted again on that machine.

For CI/CD, containers, or any non-interactive environment, set the key in the environment instead:

export AGENTVISOR_LICENSE_KEY="XXXXX-XXXXX-XXXXX-XXXXX-XXXXX"

A successful activation logs license validated ... source=online. See the Licensing guide for container and Kubernetes deployments, node-locking, air-gapped use, and troubleshooting.

Upgrading​

Re-run the install script to upgrade to the latest release:

curl -fsSL https://raw.githubusercontent.com/manetu/agentvisor/main/install.sh | sh

Next Steps​

With a license key, Docker, Python, and the CLI in place, continue to Your First Agent to scaffold, run, and test a real chatbot agent in under five minutes — then follow the rest of the Tutorial to write policies and go from development to production. When you finish, the Advanced Tutorials go deeper on content mediation, authorization enrichment, credential brokering and more.