EGL Code Accelerator for IBD

The EGL Code Accelerator is a plugin developed for IBM Business Developer (IBD) that generates responses using a Python backend application.

The Code Accelerator supports two model deployment options:

  • Fine-tuned Granite 8B Model (Local Deployment) – Requires an Ollama server and runs on the local system.

The plugin provides two main features:

  • Chat UI: Users interact with a chat interface to submit prompts and receive model-generated responses.
  • Inline Suggestions: Within the IBD editor, pressing Ctrl + Shift + Space activates inline suggestions. Suggestions appear as ghost text and propose EGL code, similar to GitHub Copilot functionality.

Prerequisites

Before using the EGL Code Accelerator, complete the following configuration steps:

Using EGL Code Accelerator

Follow these steps to start using the accelerator inside IBD.

  1. Navigate to Window 🡪 Show view 🡪 Other…

  2. Select EGL Code Accelerator under the Code Accelerator folder.

  3. The Code Accelerator UI appears.

  4. Select your LLM model from the dropdown list.

  5. Enter your prompt and start interacting with the accelerator.

Chat History

To clear chat conversations:

  • Click the Clear Chats button at the top-right corner of the UI.

Clearing chats removes them only from the UI. Chat data is still stored locally at: C:\Users\<username>\EGL Code Accelerator\IBD_Chats

Note:
  • Files are saved using the date and model name.
  • Backend logs and error reports are available in the logs folder.

Inline Suggestions

Inline suggestions enable you to generate EGL code directly within the IBD editor. Press Ctrl + Shift + Space to display context-based suggestions as ghost text. To insert the suggested code, press the Tab key, or continue typing to ignore it. If user enter a comment, that will be taken as input to generate next lines of code.

Starting the Backend Server

The backend server must be running before using the Code Accelerator.

Follow the steps below to run the backend server.

  1. Navigate to the path <RBD install path>\bin\EGL Co-Pilot

  2. Run .exe file as administrator.
  3. The server runs on port 8000.

Note:
The first startup may take some time.

Limitations

  • Currently, the EGL Code Accelerator has limited knowledge of RUI based development.
  • The current LLM remembers only the one last interaction. Full conversation memory is not yet supported.