OpenAI Forge

Automate your communication between Visual Studio Code and ChatGPT 🤖.

Screencast

https://github.com/ivangabriele/openai-forge-vsce/assets/5957876/fbb5ba36-a9b5-4251-868c-0fa6ef1676b5

Getting Started

Check the WELCOME document to get started.

How it works?

OpenAI Forge VSCode extension runs a WebSocket server while the OpenAI Forge browser extension runs a WebSocket client when you have https://chat.openai.com open in your browser. That's how VSCode can communicate with ChatGPT web application to automatically fill the prompt and submit it.

As for the errors, OpenAI Forge VSCode extension runs an Evaluator_**, that is a CLI command able to output errors while attempting to compile or lint your code. This evaluator command is run in a child process, its stderr output is extracted and embedded along your selected documents source code, before being sent to ChatGPT prompt, along with your document source code.

Features

Directly from Visual Studio Code, you can:

• [x] Send all the source code of one or multiple documents at once
• [x] Send evaluation errors to ask ChatGPT for fixes
• [x] Add a custom message to the generated prompt request (or don't)

ChatGPT Prompt Contextualization

OpenAI Forge add some contextual information to the automated prompt message in order to help improve ChatGPT accuracy:

• [x] Source code documents paths
• [x] Some workspace information "guessed" from you files path and extensions (can be disabled in settings)

Current limitations

In progress...

Known issues

In progress...