Show HN: Deep Memory – Vocabulary-driven graph memory for AI agents

Category: library

Tags: graph-memory, ai-agents, knowledge-graph, mcp, typescript

Score: 7.0/10 (Innovation: 7, Technical: 7, Documentation: 8, Utility: 6)

Deep Memory is an open-source library that provides vocabulary-driven graph memory for AI agents, enabling structured, persistent knowledge storage with schema validation. Its innovative approach uses a compact vocabulary to eliminate cold-start guesswork and token waste, while offering pluggable storage backends (SQL Server, Neo4j, CosmosDB) and MCP server integration for frictionless agent interaction. It is interesting for its practical solution to a known AI memory problem with production-ready documentation and starter kits.

Target audience: backend devs, ai engineers, data engineers

Repository: https://github.com/TjWheeler/deep-memory · TypeScript · Apache-2.0

View on Hacker News