Victor Taelin's OptMem Puts Permanent Agent Memory in a 426-Token Prompt and One Dependency-Free Python File

GitHub·high signalXBlueskyLinkedInCopy link

VictorTaelin/OptMem crossed 1,000 stars in six days (created 2026-07-25, pushed 2026-07-31) with a deliberately anti-infrastructure design: no vector DB, no embeddings, no background process. Memories are 280-byte lines in an append-only LOG.txt; a binary tree of one-line summaries (#0-1, #2-3, then #0-3, and so on) is a rebuildable cache, and merges happen on demand via `memo nap`. The install prints a 426-token markdown block you paste into AGENTS.md or CLAUDE.md and you are done. The scaling claim is the headline number: at a million memories (608 MB), `memo wake` takes 0.03s. Stated limitation — subagents cannot use memo independently.