openclaw-amemAgentic Memory for OpenClaw

Dynamic memory networks, automatic link generation, memory evolution, and hybrid BM25 + vector retrieval. No Python required.

Get Started

View on GitHub

🔄

Dynamic Memory Network

Inspired by Zettelkasten. Memories are stored as nodes in a graph, not just flat vector rows. Links are automatically generated and evolved over time.

🔗

Automatic Link Generation

New memories automatically link bidirectionally to existing related memories via embedding similarity + LLM verification.

🧬

Memory Evolution

Linked memories update context, tags, and embeddings when new details arrive. Supports active link strengthening and tag propagation.

🚦

LLM CRUD Decision Gate

Analyses user-assistant dialogue context, running NEW / UPDATE / DELETE / NONE decisions to keep memory clean and deduplicated.

🔍

2-hop BFS Graph Traversal

After vector retrieval, BFS walks the link graph up to 2 hops. Only nodes passing an embedding relevance gate are admitted, preventing noise.

🀄

Chinese-Optimized BM25

Uses Jieba (via @node-rs/jieba) for CJK word segmentation, dramatically improving recall for Chinese queries and mixed-language text.