Show HN: Mind-mem – Zero-infra agent memory with 19 MCP tools (BM25+vector+RRF) (github.com)

A new tool named mind-mem has been introduced, enabling zero-infrastructure, drop-in memory solutions for AI coding agents such as Claude Code and Codex CLI. This tool transforms agents from merely storing chat histories and notes into a comprehensive Memory OS that incorporates features like hybrid search, contradiction detection, and drift analysis. With mind-mem, multiple agents can share a persistent memory workspace, allowing decisions made in one interface to be instantly available across others, thus enhancing collaborative efficiency and reducing memory fragmentation. The significance of mind-mem for the AI/ML community lies in its governance-aware design and advanced memory capabilities. Unlike traditional memory systems that primarily handle storage and retrieval, mind-mem actively detects inconsistencies and "drift" in knowledge across agents. Its technical innovations, including co-retrieval graphs, adaptive knee cutoffs, and a governance pipeline, ensure a deterministic and auditable approach to memory management. With its lightweight installation requiring no cloud services or complex database systems, mind-mem empowers AI developers with a straightforward yet powerful tool to enhance agent memory integrity and operational efficiency.