FareedKhan-dev/langgraph-long-memory

# Technical Summary Implements a dual-layer memory architecture for LangGraph agents using thread-level (short-term) and cross-thread (long-term) storage, with long-term memory persisted as JSON documents in namespaced memory stores. Supports multiple persistence backends—InMemoryStore for development, local file-based storage with `langgraph dev`, and PostgreSQL with pgvector for production deployments with semantic search capabilities. The system integrates human-in-the-loop feedback mechanisms and checkpoint-based state management to enable agents to retain knowledge across sessions while maintaining contextual awareness within active conversations.