AI Agent Engineering > Install and configure a Secure Agent
  

Install and configure a Secure Agent

A Secure Agent is required to execute Python code, connect to MCP servers, and deploy your AI agents. Install the Secure Agent on a physical or virtual machine that runs a supported version of Linux.
To install the Secure Agent and ensure that all features of AI Agent Engineering work properly, complete the following tasks:
  1. 1Install the required libraries for Secure Agent startup if they're not already installed on the machine where the Secure Agent will run.
    2. For more information, see External library installation for Secure Agent startup.
  2. 2Download and install a Secure Agent.
    3. You can install one or more Secure Agents to run within your network or a cloud computing services environment like AWS, Google Cloud, or Microsoft Azure. You can install one Secure Agent in each user account on a physical or virtual machine.
    For more information, see Secure Agent installation.
  3. 3If your organization uses an outgoing proxy server to connect to the internet, configure the required environment variables.
    4. For more information, see Environment variable configuration for proxy servers.
  4. 4If your AI agents execute Python code and you get a missing library file error when your AI agents run, install the missing Python libraries.
    5. For more information, see External library installation for custom Python code.
When you install a Secure Agent, it's added to its own group by default. A Secure Agent group is also called a runtime environment. For more information about Secure Agents and Secure Agent groups, see Runtime Environments in the Administrator help.
You can deploy an AI agent using a runtime environment that contains the following types of Secure Agents:
You can't deploy an AI agent using the Informatica Cloud Hosted Agent, an elastic runtime environment, a serverless runtime environment, or any runtime environment that contains Secure Agents that run on Windows.