Show HN: Make0 AI Provides Shared Memory Across AI Agents

Summary

Make0 AI is a local memory system designed to give multiple AI agents, including Codex and Claude Code, access to the same user context. Its creator developed it after finding that manually handing off summaries between agents was inefficient and that a shared instruction file grew beyond 20,000 tokens, consuming too much context. The system retrieves only information relevant to each prompt, then evolved into a graph-based memory called Brain. Users can connect a Brain to compatible agents through MCP; agents can read and write memories covering preferences and other long-term context, while the user manages the data centrally. The product is designed for non-developers, with drag-and-drop memory import and a visual brain interface. The underlying system combines hybrid search with graph memory and includes built-in conflict resolution for memories. Reported LoCoMo results are 83.3% overall and 92.5% for single-hop queries in the currently available standard mode. An experimental max mode reportedly reaches 92% overall and 94.4% on single-hop queries. These figures are presented by the project creator on its benchmark page.