MemClawz
MemClawz is an open-source shared memory system for AI agent fleets, providing a collective knowledge layer that multiple agents can read from and write to simultaneously.
About
MemClawz is an open-source shared memory system for AI agent fleets, providing a collective knowledge layer that multiple agents can read from and write to simultaneously. It uses a hybrid search approach combining BM25 keyword matching with vector similarity search (weighted 70/30) to surface relevant memories, and scores results across four dimensions: semantic relevance, recency, importance, and usage frequency. Beyond simple storage and retrieval, it includes an orchestration layer that routes tasks to specialist agents using machine-learned domain classification across 13 domains.
Teams building multi-agent AI systems with OpenClaw who need agents to share knowledge and coordinate specialised work, and who have the infrastructure capability to run a self-hosted stack with Redis, Qdrant, and Neo4j.
Pros & Cons
Pros
- check MIT-licensed and fully open-source — no vendor lock-in and no ongoing licensing costs
- check Hybrid BM25 plus vector search gives better retrieval accuracy than pure semantic search alone, especially for exact-term lookups
- check Fleet Protocol v1 enables typed agent-to-agent delegation messages, making multi-agent coordination genuinely structured rather than ad hoc
- check Sleep-time reflection for pattern detection means the system improves its own memory organisation over time without manual curation
- check Simple REST API with three core operations (store, search, route) keeps integration straightforward
Cons
- close Requires Redis, Qdrant, and Neo4j as infrastructure dependencies — deploying and maintaining three separate data stores adds real operational overhead
- close No hosted or managed version available; self-hosting is the only option
- close Routing confidence scores depend on proper domain classification working correctly, which may require tuning for non-standard use cases
- close Still early-stage with limited documentation and community around it compared to more established memory solutions
More Memory
Other tools in the same category.
