Start the management-tools container in interactive mode and run the tools from within. This approach has the advantage, that it provides a familiar user experience as if the tool is installed locally.

Start a tool from the management-tools container directly, using the tool as the command argument and remove the container when the tool finishes. This approach has the advantage, that it can be used in the context of a script as it requires no interaction.

In case you need to read from or write to files on your host, you need to mount volumes or create a bind mount.

In case the containers are placed within a user-defined network, you need to add the container to that network to access the servers you want to connect. This network configuration depends on the network plugins being used. For the localhost docker-compose development setup, this network is named backend and prefixed with either the value of the COMPOSE_PROJECT environment variable or the name of the directory, where the compose files lies. In the development setup this will be compose but you can find out what networks are available, by running:

docker network list
NETWORK ID    NAME             DRIVER  SCOPE
181f1af50e6e  bridge           bridge  local
2788cdf4bd7a  compose_backend  bridge  local 
3227b513840e  compose_web      bridge  local 
494f6bdaafa8  host             host    local
3d8b08335372  none             null    local