# cognee v1.2.2 — truth-subspace reranking for the open-source agent memory platform

> Open-source agent memory gets a reranker that prefers facts the graph has already learned.

- Org: topoteretes
- Date: 2026-06-26
- Category: repo, tool
- Source: https://github.com/topoteretes/cognee/releases/tag/v1.2.2
- Canonical page: https://ai-tldr.dev/releases/topoteretes-cognee-v1-2-2/

## In short

cognee v1.2.2 adds truth-subspace reranking: an opt-in retrieval layer that builds centroids from distilled session lessons so the open-source agent memory platform reorders search hits toward what its graph has already learned.

## What it is

cognee v1.2.2 ships truth-subspace reranking on top of the open-source memory platform that gives AI agents a persistent knowledge graph across sessions. The release also hardens truth-signature hashing with sha256 and tightens centroid session filtering so the rerank stays scoped to the right context.

## How it works

Truth-subspace reranking distills past session lessons into centroids and reorders retrieval results by alignment with those centroids, with an optional learned-feedback signal you can turn on. Datasets are tracked through request-local variables so the rerank picks up the active session's context instead of bleeding across users.

## Why it matters

Vector RAG often returns relevant-but-wrong chunks because cosine similarity does not know which facts the agent has already verified. cognee's truth subspace anchors retrieval to the graph's accumulated knowledge, which is the kind of feature production RAG stacks usually have to hand-roll. Backward-compatible and off by default, so existing deployments can opt in.

## Who it's for

teams building agent stacks with persistent memory

## Try it

pip install cognee

## Links

- [v1.2.2 Release](https://github.com/topoteretes/cognee/releases/tag/v1.2.2)
- [GitHub Repo](https://github.com/topoteretes/cognee)
- [Homepage](https://www.cognee.ai)