To start this CLI in it's first iteration:

go run main.go <optional-path-to-config.yaml> <optional-path-to-data.csv>

On your first time using this application, you will be prompted for you API key which you can obtain from Open AI's website if you do not already have it to hand.

This will be saved to your own local .env file for the application to use each time it is started.

Optionally, you may add a config yaml file path as a first argument on initiation. This config yaml can also already include a field data: with the relative path to the csv data file.

The path to the csv data file may also be passed as the second argument to the initiation.

It is possible to use a csv with any number of columns. An example is provided for your use:

storage/example.csv

The prompt structure is to be provided in the configuration under a prompts: field.

An example prompt base sentence is:

prompt:

• Write me a 150 word story about a {{ .mood }} {{ .animal }} that is {{ .action }}
• Write a poem about a {{ .animal }} that you met while {{ .action }}

Each {{ .curly-braced-word }} corresponds to the column of the csv. So based on this, it is best to have a heading row and csv layout that go something like:

mood,animal,action
happy,duck,swimming
upset,bird,flying
sombre,dog,eating

Take a look at ./storage/example.csv to see examples.

The prompts will be run in the order they are provided for each row.

Each prompts output is concatenated into a single file for each row.

The output from each individual row will be written to a separate file inside a directory created and timestamped when the csv file is provided. The filename for each row's output will be row-output-# where # = the row number, so you may refer back to the supplied csv to see what prompt resulted in which section.

Sam and Madalin

Built by Vlah on behalf of Intelligiants

Proof of concept