Jervis: OpenAI chat completion CLI

A simple CLI written using Cobra and Viper to interact with OpenAI's chat completion API.

Usage

You will need to setup your OpenAI API key in your environment variable. See OpenAI API documentation for more details.

Chat command

By default, input taken from stdin, -n flag to start a new session

$ echo hello how are you | jv chat -n

Alternatively, you can type or paste in lines of text using -r flag.

$ jv chat -r

Without the -n flag, we will pick up from the last session and continue the conversation.

By default, the session are savely in the current directory in a file called .jchat.json. You can change this with the -s {session_name} flag.

Each session can be converted into a markdown file and viewed using a markdown viewer like mdcat or glow.

$ jv chat -f | glow

Edit command

The edit command only provide a convenient way to edit text. The options are the same as the chat command.

Edit command serves as a quick text editor. You can just provide the text without additional prompt.

$ cat document.md | jv edit

Code command

The code command only provide a convenient way to diagnose code. The options are the same as the chat command.

Code command serves as a quick way to diagnose code. You can just provide the code without additional prompt.

$ cat code.py | jv code

Configuration

You can configure the system prompts and other settings using the config command.

$ jv config

This would open the config file using the default editor in your terminal. The config file is store in ~/.jervis.json by default.

Installation

You can build and install locally using

$ go build -o $GOPATH/bin/jv jervis