Installation Guide

Follow this high-level sequence to get the Chucki integration running inside RAD Studio. The package installs like any other IDE extension, and you can complete the whole process in about five minutes.

Prerequisites

  • RAD Studio XE7 or newer (Delphi and/or C++Builder personality).

Step 1 — Download and install the Package

  1. Open files.chucki.ai and download the latest installer.
  2. Execute the installer on your machine and follow the instructions.
  3. The Chucki package will be installed in the automatically detected RAD Studio version.

Tip: Close the IDE before executing the installer or restart the IDE after the installer is finished.

Step 2 — Provide Login Credentials

  1. Open RAD Studio.
  2. Open Tools -> Options -> Chucki -> General.
  3. Under the License tab login with your Email and License code credentials, that we have send you via Email.

Step 3 — Provide Model Login Credentials

  1. For inline completions, open Tools -> Options -> Chucki -> Completions and sign in to your preferred provider (GitHub Copilot, local models, etc.).
  2. For Agent Chat, open Tools -> Options -> Chucki -> Chat and sign in to your preferred providers (GitHub Copilot, OpenAI, Google Gemini, or local models).

For detailed instructions please visit Settings.

Step 4 — Smoke-Test Inside a Project

  1. Open any RAD Studio project.
  2. Ensure the Chucki - Completions dockable window is visible (View -> Chucki - Completions).
  3. Ensure the Chucki - Chat dockable window is visible (View -> Chucki - Chat).
  4. Start typing in the code editor and click Ctrl + Space at a line. You should see either a Ghost text or a Chucki symbol at the top of the symbol list. If no suggestion appears, it could be possible that the AI simply does not have a suggestion in this line. Try in a few other lines in this case.
  5. Add a chat with a model of your choice (one where you are signed in) in the chat window.
  6. Send a test prompt in the chat window (for example, “Summarize the class TMyClass in @Unit1.pas”). Confirm the response uses your chosen model.
PreviousOverview NextGetting Started