I just built something similar, specific to Claude Code. It runs as a transparent HTTP proxy that reads & rewrites the entire messages array that CC sends to its API. Same “dreaming” consolidation approach (using Haiku and another instance of CC itself, so it uses your subscription). Check it out!

https://github.com/LabLeaks/spotless

Been saying this for a while. Vector similarity is the wrong primitive for agent memory. It finds things that sound related, not things that are actually relevant given current context and what the agent already knows.

The "ditching vector databases" framing is a bit dramatic since you still need embeddings somewhere, but the point stands that a raw vector store with no resolution layer is basically a pile of notes your agent can't reason about. You need conflict detection, certainty weighting, temporal scoping. Otherwise you're just building a fancier version of the same problem

I wonder if this is Google's answer to Beads?

Your vector database should be in memory