memweave: Zero-Infra AI Agent Memory with Markdown and SQLite — No Vector Database Required

LLM agents are stateless, and common memory solutions like vector databases introduce opacity, infrastructure overhead, and lack version control. The memweave library offers an alternative by using Markdown files as a human-readable source of truth for agent memory. These files are indexed into a local SQLite database, which can be rebuilt at any time, separating storage from search. This file-based approach enables version control with tools like Git and provides full provenance for retrieved memories without the complexity of a dedicated vector database service.