Use ES6 class syntax to enable interfacing in these modules.

Lodash would improve speed and give better functionality.

Roll back on using the csv module for parsing and stringifying the csv data. Also try to implement transformers to enable infinite size inputs.

This would help to give commands and configs via an exposed end point. One may use ExpressJS or SenecaJS to achieve this.
Seneca would be a great choice though.

One can use JSDoc.

One might need to have the feature of reading from the SQL tables if necessary, and output the required teamScores.csv.
One way to do this is to provide arguments to input module to instruct it to either read from SQL database or from files

Right now the files(paths) are being provided as command line arguments using caporal.js
Typical scenario when this might be use as a pluggable for AutolabJS, it would be necessary to read the details from a config.json file. So one would want to enable this feature by having a config.json file.

Modify output.js to do the same.
Can you take this up @aakp10

The default logger logs into the console. The implementation of the logger, using winston, would write logs into log files defined in the configs.