The AI Agents in WorkflowFirst applications use a specific helper tool that gives them access to a browser. This helper tool is available to download from http://myworkflowfirst.com/installs/cfagentbrowser.zip. It requires that .Net Core is installed on the server (it usually is pre-installed).

The cfagentbrowser.exe file should be placed into its own folder (preferably under Program Files).

You should run the tool to install it as a Windows service (see below).

Please note the following available command-line tools (when you run from a cmd.exe or PowerShell prompt):

Usage:
  cfagentbrowser [options]
  cfagentbrowser service <command> [options]

Options:
  -?, -h, --help                 Show this help.
  --http [port]                  Run HTTP mode. Defaults to port 8876.
  --http=<port>                  Run HTTP mode on the specified port.
  --alt-debug [on|off|status]    Enable, disable, or show alternative snapshot debug logging.
  --alt-debug=<on|off|status>    Enable, disable, or show alternative snapshot debug logging.
  --temp-folder <path>           Set the root folder used for session downloads.
  --temp-folder=<path>           Set the root folder used for session downloads.

Service commands:
  service install [--port <port>] [--name <service-name>] [--service-exe <path>] [--start]
  service uninstall [--name <service-name>]
  service start|stop|restart|status [--name <service-name>]

By running cfagentbrowser service install you will tell the system to keep this running every time Windows starts up, regardless of whether a user is logged-in. Once installed you should start the service using cfagentbrowser service start (or status to check the status of the service). Alternatively go to Task Manager / Services, find cfagentbrowser in the list and run it there.

Installing Google Chrome

The agent browser is a headless browser, meaning it uses the regular web browser but without a visible display. It is optimized to work best with Google Chrome, so it is recommended to ensure that Google Chrome is installed on the server where your WorkflowFirst application is running before using your AI agents.