# SQLite Memory

A SQLite extension that gives AI agents persistent, searchable memory over markdown files, combining vector similarity search with full-text search and offline sync between agents.

SQLite Memory is a SQLite extension that gives AI agents persistent, searchable memory optimized for markdown content. It combines hybrid semantic search, vector similarity plus FTS5 full-text search, with markdown-aware chunking and local embedding through llama.cpp.

Memory databases can synchronize between agents using offline-first sync, so each agent works independently and merges changes when it reconnects, which suits distributed agent systems and edge deployments. Markdown files serve as the human-readable, version-controllable source of truth, with embeddings added for semantic retrieval.

Because it runs as a SQLite extension, an application gets a single portable database file and a SQL interface, with no separate vector database server to run.

Website: https://sqlite.ai