Agentic Memory
What it is
A Python library providing a graph-based memory layer for AI agents, featuring a 3-tier architecture (core, recall, archival) and a custom graph engine for relationship-aware retrieval. It includes a 'sleep-time compute' mechanism that automatically consolidates, clusters, and prunes memories in the background to optimize agent context.
The problem it solves
Standard vector databases lack relationship awareness, and manual context management for long-running agents is inefficient and prone to context window overflow.
What you could use it for
- Building stateful AI agents that need long-term persistent memory
- Implementing semantic search with graph context expansion for richer retrieval
- Automating background memory consolidation and pruning to manage context window limits
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.