SuperLocalMemory: Privacy-Preserving Multi-Agent Memory with Bayesian Trust Defense Against Memory Poisoning

SuperLocalMemory is a new privacy-preserving memory system for multi-agent AI that defends against memory poisoning attacks through local-first architecture and Bayesian trust scoring. The open-source system eliminates cloud dependencies while providing personalized retrieval through adaptive learning-to-rank, demonstrating strong performance metrics including 10.6ms search latency and 72% trust degradation for sleeper attacks.

• →SuperLocalMemory addresses OWASP ASI06 memory poisoning threats through architectural isolation and Bayesian trust scoring without requiring cloud dependencies.
• →The system combines SQLite storage with FTS5 search, Leiden clustering, and adaptive re-ranking that learns user preferences through behavioral analysis.
• →Performance testing shows 10.6ms median search latency, zero concurrency errors under 10 simultaneous agents, and 104% improvement in NDCG@5 with adaptive re-ranking.
• →The system includes GDPR Article 17 erasure support and isolates behavioral data in separate databases for privacy protection.
• →SuperLocalMemory is open-source under MIT license and integrates with 17+ development tools via Model Context Protocol.