---
title: "Mwe-MCP: Self-hosted Memory for AI Agents with Multi-user ACL"
source: "Hacker News 热门（Algolia · 中文化）"
source_tier: "t2"
url: "https://github.com/Fr4nZ82/mwe-mcp"
permalink: "https://feed.astralor.com/items/itm_36ecc4fec6c7dbace5c35df1"
published: "2026-07-23T12:38:20.000Z"
score: 66
channel: "tech"
tags: ["AI Agent","Markdown Wiki","MCP","Mwe-MCP","Self-hosted"]
---

# Mwe-MCP: Self-hosted Memory for AI Agents with Multi-user ACL

## AI commentary

> Mwe-MCP cleverly decouples agent and memory storage via MCP, enabling flexible and auditable knowledge management. For developers building multi-agent systems or requiring fine-grained access control, this open-source tool is worth exploring. Self-hosted and API-free, it's especially suited for enterprise intranets or privacy-sensitive scenarios.

## At a glance

Mwe-MCP is a self-hosted persistent memory system for AI agents, using human-readable Markdown wiki format with multi-user and per-fragment ACL. It self-organizes overnight and is agent-agnostic via the MCP protocol. Open-sourced under AGPL, ideal for teams needing private, customizable knowledge management.

## Key points

- Human-readable Markdown-based memory for easy review and editing
- Multi-user with per-fragment access control
- Overnight auto-organization to enhance retrieval
- Agent-agnostic via MCP protocol, works with any AI agent framework

## Full text

Persistent, structured memory for AI agents — a human-readable Markdown wiki, multi-user with per-fragment ACL, self-organizing overnight. Agent-agnostic via MCP. Self-hosted, AGPL. - Fr4nZ82/mwe-mcp

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_36ecc4fec6c7dbace5c35df1