Note: This repository is a work in progress (WIP) and may not be fully functional yet. Use at your own risk.

Welcome to the README for the Dual Pane Brain extension.

Introduction

Dual Pane Brain is a WIP dual-pane editor for interacting with openai-compatible Large Language Model (LLM) endpoints. It is designed to provide an effortless way to create new LLM-generated content.

It is developed against oobabooga but should work with anything that presents an openai-compatiable api (including openai itself).

Upcoming features:

• Various settings toggle buttons somehow
• Better management of api instances - expect breaking changes at some point

Setup

To get started with Dual Pane Brain, follow these steps:

1. Install the extension from the Visual Studio Code Marketplace.
2. Run the command >dpb.setupApiInstance.
3. Fill out the prompts:
   • a display name for this instance
   • the model this instance is running
   • your preferred openai-compatible endpoint URL (e.g., http://oobabooga.local:5000, https://api.openai.com/v1/)
   • your api key

Generating

To generate, open a text document in vscode.ViewColumn.One (the left hand side). We'll refer to this editor as the prompt pane.

Type up a prompt as desired:

How do I bake cookies?

Once you have your prompt ready, click the Generate button located on the right side of the status bar.

This will send the contents of the prompt pane to the AI API as a user message. A new document will open in vscode.ViewColumn.Two (response pane), and API responses will be streamed and displayed there.

Continuing

This is where the magic happens.

The Continue button does a few different things based on the selected text in both the prompt and response panes.

When no text is selected in either pane, clicking the Continue button takes the contents of the prompt pane and response pane before the cursor and sends it to the AI API, appending the newly generated text after the cursor.

When text is selected in either pane, the selected text is sent to the AI API as the prompt. The generated response is inserted after the cursor.

Troubleshooting

If things seem off, try deleting everything in your settings.json that's namespaced under dpb. Then, reload the window, re-add your instance, and try again.

Instance management is a high priority on the development roadmap.

Wrapping up

Contributions are welcome! If you have any ideas or suggestions, please submit an issue or pull request on the GitHub repository.