Running the agent in docker is the closest thing to this that currently exists, and it certainly does the job for someone who's getting started.
Roughly, the process is:
- Install a linux distro (virtual or physical) that supports a recent version of docker, install docker and git on it
git clone https://github.com/openai/universe.git then
cd universe to clone the repo and enter the directory
docker build -t universe . to create a docker image tagged as 'universe'
- Follow the instructions at https://github.com/openai/universe/blob/master/doc/remotes.rst#agents-inside-docker, replacing the image name 'my/agent:latest' with 'universe'
Note that you can use the --it parameter and add /bin/sh to the end of the 'docker run' command to open a shell, then you can run 'python' and follow the instructions in the main readme to experiment.