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

2026-04-16 · Source: Towards Data Science · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

memweave is an open-source project designed to provide persistent, auditable memory for LLM agents, addressing the limitations of traditional context stuffing and vector databases. It stores memories as Markdown files on disk, indexing them into a local SQLite database for hybrid BM25 + semantic vector search. This approach allows for version control via `git diff`, direct inspection of memory files, and eliminates infrastructure overhead associated with vector stores. memweave's architecture separates storage from search, using Markdown files as the source of truth and the SQLite database as a derived, rebuildable index. It incorporates a deterministic write path with SHA-256 hashing for efficient caching of embeddings and a search path that merges BM25 and vector search results, followed by optional temporal decay and MMR re-ranking to prioritize recent, diverse, and relevant information.

Key takeaway

For AI Engineers building LLM agents that require persistent, auditable, and context-aware memory, memweave offers a lightweight, file-based solution. You should consider integrating memweave to avoid stale context issues and infrastructure overhead, allowing your agents to retain and reason about knowledge over time with full transparency and version control.

Key insights

LLM agent memory can be managed effectively using version-controlled Markdown files and a local SQLite index.

Principles

• Separate storage from search for agent memory.
• Files as source of truth enable auditable memory.
• Hybrid search improves retrieval accuracy.

Method

memweave processes Markdown files into a SQLite index using SHA-256 hashing for caching, then performs hybrid BM25 and vector search, applying optional temporal decay and MMR re-ranking for optimal results.

In practice

• Use `git diff` to audit agent memory changes.
• Configure `decay_half_life_days` to manage memory freshness.
• Employ `source_filter` for multi-agent memory isolation.

Topics

• memweave
• AI Agent Memory
• Markdown Files
• SQLite Database
• Hybrid Search

Code references

• sachinsharma9780/memweave

Best for: AI Engineer, Machine Learning Engineer, AI Architect

Related on AIssential

• A 0.12% parameter add-on gives AI agents the working memory RAG can't — Delta-mem offers an efficient, dynamically updated associative memory for AI agents, overcoming RAG and context window limitations.
• MemRefine: LLM-Guided Compression for Long-Term Agent Memory — LLM-guided compression, MemRefine, manages agent memory by using an LLM judge for factual content-based decisions, outperforming surface similarity.
• A semantic memory layer for local AI agents — no vector DB, one file, runs on CPU — A lightweight, local semantic memory solution for AI agents avoids complex vector databases for smaller datasets.
• AI Agent Amnesia? Here’s the Open-Source Fix That Works. — Hermes Agent uses SQLite FTS5 and LLM summarization for persistent, cross-session AI memory.
• Building Cost-Efficient Agentic RAG on Long-Text Documents in SQL Tables — An Agentic RAG architecture can integrate LLMs with existing SQL databases for hybrid structured and semantic queries.
• EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective — EvoMemBench systematically evaluates LLM agent memory across scope and content, revealing current limitations and specialized strengths.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Towards Data Science.