Introduction

Every LLM has a hard limit on how much it can see at once — and every token costs money. The bigger your project, the longer your session, the faster you hit that wall. Most AI coding tools either crash into it or blindly stuff the model with everything and burn through your budget in minutes.

Umbra is built around a different approach: aggressive, layered context management that keeps the model effective within a fixed budget, no matter how large the project or how long the session runs.