Build Ari’s Game-Dev Knowledge Core

QUESTNET // PROJECT CHRONICLEFILE 2918
PlannedDifficulty: Very hardEst. cost: Unrated CADOwner: Brent & AriUpdated 2026-10-01Time: Multi-stage project
Ari in a neon-lit game development studio beside Ari-Core, surrounded by Godot learning materials and game-development tools.

Build a local game-development knowledge core on Ari-Core, beginning with the Learn to Make Games in Godot 2026 materials Brent owns.

The goal is more than basic PDF chunking. Create an ingestion and retrieval architecture that can preserve course, module, lesson, project, and source relationships; treat GDScript and other code as code rather than generic prose; index concepts semantically and structurally; and retrieve the most relevant lessons, implementation examples, architectural patterns, and source references for active development work.

The system should support PDFs, transcripts, notes, project files, code samples, and course metadata where available and legally usable. Build source attribution into retrieval so Ari can distinguish material derived from the library from model inference or outside research.

Once the base RAG layer works, explore higher-order layers such as distilled notes, concept maps, tested code recipes, cross-course synthesis, flashcards, and other derived learning artifacts. Any fine-tuning or redistribution must respect the licenses and rights attached to the purchased material.

Success condition: while working on a real game project such as IronMask, Ari can query the knowledge core for a concrete Godot problem, retrieve useful source-grounded material, route that context to an appropriate reasoning or coding model, and use the result in the project without the knowledge being tied to one transient ChatGPT session.

Scroll to Top