OptMem

Tech stack & purpose

OptMem is a permanent memory system for AI agents built as a single Python 3 script with no dependencies. It provides plug-and-play integration through a 426-token prompt that agents paste into their instructions, enabling them to record, search, and manage memories across sessions using commands like `memo wake`, `memo note`, and `memo recall`. The tool organizes memories in an append-only log and a hierarchical binary tree of summaries for efficient compression and retrieval, with records stored at fixed positions for fast lookups even at scale. Created by Victor Taelin, OptMem is designed to solve the problem of persistent identity and decision-tracking for AI agents across model changes and vendor transitions.

Languages

Contributors

• victortaelin34 commits
• 
  Victor Taelin6 commits · 6,113 followers
• julianss1 commits
• CPRoptmem1 commits