Memory offers a persistent, SQLite-backed storage solution designed for agents, providing both a command-line interface and a programmatic API. It ensures reliable memory management with broad cross-platform support by leveraging `sqlite3`, avoiding complex native build steps. Key features include optional FTS5 indexing for enhanced search capabilities, robust input validation to prevent oversized data payloads, and automatic semantic embedding generation via OpenAI for advanced contextual recall. This allows agents to efficiently store, recall, list, forget, export, and import memories, with fallbacks to text-only scoring when embedding keys are not provided.
Key Features
01CLI and programmatic API for flexible integration
02Robust input validation and data limits for memory payloads
030 GitHub stars
04Optional FTS5 indexing for enhanced search capabilities
05SQLite-backed persistent memory storage
06Automatic semantic embedding generation via OpenAI
Use Cases
01Managing agent-specific data like preferences, facts, or historical interactions
02Building applications that require semantic search and retrieval of stored content
03Providing persistent memory for AI agents to store and recall information
