Show HN: TenseMem – deterministic memory for AI agents

Category: ai-ml

Tags: ai-agents, memory, sqlite, llm, deterministic, context-management

Score: 7.3/10 (Innovation: 7, Technical: 7, Documentation: 8, Utility: 7)

TenseMem is a deterministic memory layer for AI agents that maintains one current truth per predicate slot using an append-only episode log projected into SQLite. It distinguishes current vs. stale facts, tracks freshness per read, fences recalled context to prevent self-echo contamination, and supports reversible corrections. The model proposes typed operations while the engine deterministically applies them, giving agents memory they can safely act on.

Target audience: AI engineers building LLM agents and applications requiring persistent, reliable conversational memory

Repository: https://github.com/tensemem/tensemem · Python · Apache-2.0 · 3 stars

View on Hacker News