A simple ChatGPT powered chat for PowerShell.
https://github.com/Ross-Murphy/PowerShell-GPT
Simply import the PowerShell module and you can call up a simple command line chat interface with OpenAI ChatGPT to help with whatever you like. Add the module command to your PowerShell profile and you will always have it ready to go.
This started off as a way to have a simple coding and virtual assistant inside VSCode or in my other PowerShell task windows. I did not initially find an extension for VsCode to my liking. Also I thought it would be fun to code something like this and it only took a few hours of innovation time.
There are several other PowerShell OpenAI ChatGPT API scripts out there and honestly if I had found them first I may not have started this at all, however it was fun and I like the simplicity of it all. This needs to become a proper module if it's to be distributed as a project. Eventually I would like to add some ability for the chat to interact with the shell in a limited capacity. Using the new Functions array in released in gpt-3.5-turbo-0613.
Setup and Config Handling
-
Reading config file will detect version mismatch and force setup wizard.
-
Change Setup function to allow accepting default or current options making in quicker to change settings without more input.
-
Changed the setup and config process to allow changing between model with a menu. For example gpt-3.5-turbo-16k has 4 times the chat context size however prices per token are higher as well. This somewhat addresses Issue: #1 Where longer chats or large conversational context causing an error.
-
Add Debugging option to Global Config . The ability to turn on extra messages like debugging that will show token count and other verbose output. use
Setup()to change setup options
Added new menu command aliases
-
Q()for save and quit. Now you can drop back to the shell and then when back in chat import the saved chat withI()and pick up where you left off with history context. -
M()shorthand forMulti()Multiline input mode. Default input mode is press <enter> to send. However if you want send multiple lines, like pasting in some code to analyze use M() to activate multiline mode. Put a period on a line by itself to send, like this:. <enter>
v0.5.3
-
Previous version of config file causes upgrade setup failure
-
Added support for OpenAI GPT-4 models 'gpt-4' & gpt-4-32k
Supported models are now
- 'gpt-3.5-turbo' # Default
- 'gpt-3.5-turbo-16k'
- 'gpt-4'
- 'gpt-4-32k'
-
Typo in model selection menu causing error
-
Issue #2 Non ascii characters returned by the API would cause an error if re-submitted by the message array with out proper handling.
Get an API key from your OpenAI Account https://platform.openai.com/account/api-keys
Clone the repository
git-clone https://github.com/Ross-Murphy/PowerShell-GPT.git
Import Module.
Import-Module -Name \path-to\PowerShell-GPT\PowerShell-GPT.psm1
Run Start-Chat to kick off the setup for first time.
As of version 0.4 this should run anywhere that runs PowerShell 7.x on Linux or Windows etc.
Adding the module import command to your PowerShell Profile is handy.
In PS terminal window substitute code for your preferred editor.
code $PROFILE
In this example below you can add it to your profile and use it in VsCode
In the vscode PS terminal window
code $PROFILE
add the following adjusting for your own path to your location where you git cloned the repo.
Import-Module -Name /path-to/PowerShell-GPT.psm1 -ForceWhen you next login and the module is loaded you simply run start-chat to get started.
start-chat
You are now chatting with gpt-3.5-turbo. Type your chat message and hit <enter> to send.
Or choose a command from the menu.
---
GPT-PowerShell Version 0.5
==================================================================================
Name Command Description
==================================================================================
Close Chat: Quit() or Exit() End chat session. Alias Q() is Save and Quit.
Multiline input mode: Multi() or M() Multiline text entry mode, Use Dot-escape to exit .<enter>
Save/Export Current Chat: Save() or S() Export Contents of current chat messages to $Env:GPT_CHAT_MESSAGES
Import Saved Chat: Import() or I() Import content of $Env:GPT_CHAT_MESSAGES & append to current messages array.
Reset Chat Session: Reset() Clear messages array. Start fresh chat.
History: History() Display Chat History. See Content of current messages array.
Config: Conf() Display Current Configuration.
Setup: Setup() Setup config options. API-Key, system_msg, model context, debug msg
Help: Help() Display this help menu.
: Good Day.
Good day, Sir. How may I be of assistance to you?
: quit()
Run start-chat Note: The first time it will run Setup(). You will get prompted to configure your API key.
If you want to run the setup again at any time you can always run Start-PowerShellGPTSetup in PowerShell
Or alternatively you can use Setup() from the chat command menu.
The setup creates a configuration directory in user home dir called .PowerShell-GPT and stores and loads the config in a json file stored there.
https://platform.openai.com/docs/introduction
https://platform.openai.com/docs/api-reference
'Copyright (c) 2023 Ross Murphy - MIT License'
TLDR. - A short, permissive software license. Basically, you can do whatever you want as long as you include the original copyright and license notice in any copy of the software/source.