🧠 Augmented Context (Codebase)

The Augmented Context feature allows CodeGPT to "index" your entire project. This creates a local vector database of your code, enabling the AI to retrieve relevant snippets from across different files to provide much more accurate and context-aware answers.

🔍 Why Index your Codebase?

Standard LLMs only see the file you have open. By indexing your workspace, CodeGPT can:

  • Find existing logic: Ask "Where is the authentication logic handled?" and get the exact file and line.
  • Refactor safely: Understand how changing a function in one file might affect others.
  • Learn patterns: The AI captures your project's unique coding style and architectural patterns.

⚙️ How to Enable Codebase Indexing

  1. Open the CodeGPT Chat panel.
  2. Click on the "Augmented Context" button or the "Codebase" icon.
  3. Click "Select Folder to Index".
  4. Once chosen, CodeGPT will start scanning your files and creating a local index.
Augmented Context panel showing Codebase indexing with Select Folder to Index button

Note: The indexing process is local and private. Your code is not sent to our servers for indexing.

🚀 Using Indexed Context

Once indexed, you can use the @codebase mention in your chat prompts to tell the AI to search your entire project for the answer.

"Analyze @codebase and explain how the data flow works for the user registration process."

🛠️ Management

  • Exclude Files: You can use a .codegptignore file to prevent specific folders (like node_modules) from being indexed.
  • Refresh Index: If you make major changes to your project structure, click "Refresh Index" to keep the AI updated.

🔜 Next Step

Explore the different Interaction Modes (Chat, Search, and Agent).