The Python Agent discovers, and maps and tracks metrics for business transactions, app services, and backends in your web application by injecting instrumentation into the Python application at runtime. The agent operates from inside the WSGI server that contains the instrumented application.

The agent proxy is a Java process that handles the communication between the Python Agent and Controller. The proxy reports performance metrics to the Controller, where the data is stored, analyzed, and presented.

As shown in the following figure, the proxy can serve multiple Python Agents simultaneously.

In pure Python environments, the proxy is automatically started when you start the Python Agent. In other types of environments, you need to start the proxy manually.

However the proxy is started, the commands that start the agent check whether the proxy is running before attempting to start it.