sumanthreddyy/Smriti

Smriti is a brand-new project (2 days old) entering the crowded 'AI Memory' market. While it correctly identifies the current technical trend of combining Vector RAG with Knowledge Graphs (GraphRAG) and BM25 for robustness, its zero-star status and lack of community traction put it in the 'personal project' category. The inclusion of an MCP (Model Context Protocol) server is a smart strategic move to leverage Anthropic's ecosystem, but this also places it directly in the crosshairs of frontier labs. OpenAI and Anthropic are both aggressively building native long-term memory and context management into their platforms (e.g., OpenAI's Memory feature and Claude's Project Knowledge). Competitors like MemGPT, Zep, and LangGraph's checkpointing already have substantial lead times, developer mindshare, and production hardening. The 'offline-first' angle provides a minor niche for privacy-focused users, but without a significant architectural breakthrough or massive data gravity, it is likely to be superseded by platform-native capabilities within 6 months.