Show HN: Memharness – Bi-temporal memory for AI agents, in one SQLite file

Category: library

Tags: ai-agents, memory, sqlite, bi-temporal, mcp, typescript

Score: 7.8/10 (Innovation: 8, Technical: 7, Documentation: 9, Utility: 7)

Memharness is a bi-temporal, provenance-tracking memory layer for AI agents, stored in a single SQLite file with no LLM calls. It uniquely combines temporal versioning, supersession-based corrections, and per-fact provenance into a deterministic, embeddable library exposed via MCP tools, making agent memory auditable and queryable over time.

Target audience: backend devs, ai engineers, data engineers

Repository: https://github.com/las7/memharness · TypeScript · Apache-2.0 · 1 stars

View on Hacker News