Causal graph memory for LLMs. Flat token cost, no matter how the session runs

(github.com) by w4mwati | Jun 19, 2026 | 0 comments on HN