Clio-Memory

Clio-Memory

A self-hosted MCP memory and state system for long-running AI relationships, storing source text locally with explicit memory tools and a web manager for humans.

Category
访问服务器

README

Clio Memory

Clio Memory is a self-hosted MCP memory and state system for long-running AI relationships and assistants. It stores source text locally, exposes explicit memory tools, and provides a warm web manager for humans.

Clio is named after the Muse of history in Greek mythology: not because memory should trap a person in the past, but because continuity gives later conversations somewhere real to begin.

This repository is a privacy-clean source shell. It contains no user memory, API key, domain, tunnel credential, manager password, push key, database, log, backup, or private persona.

What it includes

  • Semantic and keyword hybrid retrieval with local bge-small-zh-v1.5 embeddings
  • Startup memory index with fixed and flowing layers
  • Exact source recall with pagination and newest-first reading
  • Append-only memory writes, write-before snapshots, history retention, and safe split tools
  • Independent mailbox with search, edit, delete, history, and retention
  • Feeling memories and V/A mood resonance
  • Prospective reminders and date-based memory calendar
  • Fact timeline with automatic candidate detection and human confirmation
  • Contradiction warnings without silent overwrite
  • Sealed memory, pin levels, topic cabinet, relations, and retrieval feedback
  • Unfinished matters, AI treasury, private thoughts, resonance, tension, and state causality
  • Silence/static state engine, optional Bark actions, and a private editable judge configuration
  • Dry-run digestion planning that never archives or merges automatically
  • Authenticated responsive web manager and encrypted exports
  • Response seal and request diagnostics

See Features for the complete tool list.

Quick start

Requirements: Docker Desktop on Windows, or Docker Engine with Compose on Linux.

Windows

git clone https://github.com/dan521627-hash/Clio-Memory.git
cd Clio-Memory
powershell -ExecutionPolicy Bypass -File .\start.ps1

Linux / VPS

git clone https://github.com/dan521627-hash/Clio-Memory.git
cd Clio-Memory
chmod +x start.sh
./start.sh

The first start creates a private .env, an editable config.yaml, and empty local data directories.

  • MCP endpoint: http://127.0.0.1:18001/mcp
  • Web manager: http://127.0.0.1:8787
  • Health check: http://127.0.0.1:18001/health

The default Compose file binds both ports to localhost. It does not create a public tunnel or fixed domain. Public access requires the deployer's own HTTPS reverse proxy and security configuration.

Optional model and push services

The service can run without an API key. When an OpenAI-compatible API is configured, it can improve summaries, emotion evaluation, task extraction, topic suggestions, and fact-change candidates. Put the deployer's own key in .env; never paste it into source files.

Bark delivery is optional and disabled until the deployer supplies a device key and changes behavior mode from rehearsal after testing.

Data ownership

Runtime data lives under ./data, model files under ./models, and exports under ./exports. These directories are ignored by Git. Back them up separately. Never publish them with a fork.

Before exposing the manager or MCP endpoint outside localhost, read Privacy and security.

Project status

This is an experimental personal-memory system, not a medical device or a claim that language models possess biological hormones or human consciousness. State names are computational metaphors used to create continuity and inspectable behavior.

Attribution and license

Clio Memory is based on P0lar1zzZ/Ombre-Brain, released under the MIT License. The original license is preserved in LICENSE. Modifications remain available under the same license.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选