Index Codebase (The Analytical Mind)

One of the most valuable points of Gamedev AI, differentiating it drastically from conventional ChatGPT or other VSCode co-pilots, is its deep knowledge of your game project's connective mesh.

If "Open Context" is the AI's current eye, Index Codebase is the robot's dormant neural web.

What is Vector Indexing (Vector DB)?

When working with immense frameworks like Godot, many custom classes like PlayerStats, InventoryManager, or WeaponBuilder do not belong to the classic GDScript manual the AI studied years ago at OpenAI. They are classes you coined.

Going there and pasting the PlayerStats script every time in the chat to ask about inventory devours thousands of repetitive tokens and your time. By embedding DB Indexing (Vector Search), this ends.

Operating the Indexing (Settings Button)

1. Go to the Gamedev AI panel and click the hidden gear tab for Context options ("⚙️ Settings" in the bottom corner).
2. Scroll down the options panel to reveal the Vector Database window.
3. Click on 🔍 Scan Changes. The plugin will sweep all your File Names on disk, looking for unreported deletions or edits.
4. The visual File List will populate in colors pointing out crucial additions and old unchanged ones. If there is a discrepancy from the previous final Indexing:
5. Finally, click ⚡ Index Database.

This will activate the invisible indexer mode with Gemini's text-embedding API. At the end of the progress hourglass, a semantic JSON RAG network will be hidden and encrypted on disk called .gamedev_ai/vector_db.json.

Now, try in the chat:

• "How do I fix the inventory bug?"

The AI will respond: "Oh, I noticed from your scripts in res://src/player_inventory.gd and the call in Network.gd that the classes are not persistently connected. I'll adjust that."

Next Step: Understand the 25 "Skills" Files that Help RAG \u27A4