Memory Bank

Vector-native AI agent memory with multi-backend storage, shared spaces with ACL controls, and semantic search

protocol-lattice 14
Claude CodeClaude DesktopGeneric
View source ↗

Production-ready MCP server providing vector-native memory capabilities for AI agents with support for multiple database backends including PostgreSQL with pgvector, Qdrant, MongoDB Atlas Vector Search, and in-memory storage. Features two-tier memory architecture with short-term session buffers and long-term vector storage, shared memory spaces with fine-grained ACL controls and TTL support, and dynamic embedding configuration through AutoEmbedder supporting OpenAI, Gemini, and local models. Exposes comprehensive tools for memory management, contextual retrieval, collaborative spaces, and performance monitoring.

Source

Repository: https://github.com/protocol-lattice/memory-bank-mcp

Maintain Memory Bank?

Let people know it's listed here — add the badge (live metrics, light/dark aware) or a plain link to your README or docs.

[Memory Bank on getagentictools](https://getagentictools.com/mcp/protocol-lattice-memory-bank-mcp?ref=badge)