Token-Efficient Memory: How Aura Caches BM25, Repo Maps, and Long-Term Context

Most AI coding tools begin each session by re-reading your entire project. They re-index the same files. They resend the same system prompts. And they have no memory of what you did yesterday. Aura ta