Engram is a production-ready memory system that gives AI agents persistent, searchable memory across conversations. Think of it as SQLite for agent state - simple, fast, and reliable.
Features
🧠 Hybrid Memory Retrieval - Combines semantic similarity, recency, confidence, and access patterns
🔍 Full-Text Search - SQLite FTS5 for fast keyword search
🎯 Vector Embeddings - Local embeddings with Xenova/all-MiniLM-L6-v2 (~23MB)
🔐 Secret Detection - Automatic detection and redaction of API keys and credentials
🧹 Auto-Consolidation - Duplicate detection, contradiction flagging, confidence decay
📊 Web Dashboard - Modern React UI for visualizing and managing memories
🔌 MCP Server - Model Context Protocol integration for Claude Desktop/Code
🚀 REST API - Full-featured HTTP API with CORS support
📦 Zero Config - Works out of the box with sensible defaults