ZR

zackify/revect

开发工具

本页由 AI 翻译,技术细节请以原文为准。 查看原文

20 stars 1 forks 质量 54 趋势 10

[ALPHA]支持在任何AI工具上持久化和回忆信息,支持SQLite + MCP + 本地嵌入

概览

🎬 演示视频即将上线!请见下方示例的使用截图。🎬 该项目目前处于测试阶段。功能和界面可能会发生变化。 **call **ors 是你的个人记忆库🔒——一个自托管的工具,可以无限期地保存和回忆任何信息。永远不要再丢失宝贵的知识! 最初是为你过去读过但记不清确切标题或内容而设计的。你知道你读过相关内容,但记不清细节?Revect 通过语义搜索帮你找到它!📚 但我们意识到它对 非常有用。无论你是单独工作还是与他人协作,revect 都成为你的集体知识库,任何 AI 都能利用它。🤝 有了MCP支持,你可以用revect作为私密方式拥有你的数据,并在任何AI系统中无缝调用数据。你的数据,你的控制权。🛡️ 🔄 :任何嵌入模型,大小不一——从微小高效模型到巨型强力模型。

概述(AI 翻译)

🎬 演示视频即将上线!请见下方示例的使用截图。🎬

注: 本项目目前处于测试阶段。功能和界面可能会有所变化。

Recall vectors 是你的个人记忆库🔒——一个自托管工具,可以无限期地保存和回忆任何信息。永远不要再丢失宝贵的知识!

🎯 这到底是做什么用的?

最初是为**回忆过去读过但不记得确切标题或内容的文章和博客而设计的。你知道自己读过相关内容,但细节却记不清?Revect 通过语义搜索帮你找到它!📚

但我们意识到,它对于团队和个人在众多AI工具中共享且持久的记忆非常有用。无论你是单独工作还是与他人协作,revect都成为你任何AI都能利用的集体知识库。🤝

有了MCP支持,你可以用revect作为私密方式拥有你的数据,并在任何AI系统中无缝调用数据。你的数据,你的控制权。🛡️

🔄 嵌入模型自由:带任何嵌入模型,大小不一——从微小高效的模型到庞大的强大模型。

本页由 AI 翻译,技术细节请以原文为准。

README

以下为原文 README 节选,完整内容请查看原文。

revect 🚀 ✨ 🧠

🎬 Demo video coming soon! See Examples below for usage screenshots. 🎬

Note: This project is currently in alpha release. Features and interfaces may change.

Recall vectors is your personal memory vault 🔒 - a self-hosted tool to persist and recall any information indefinitely. Never lose valuable knowledge again!

🎯 What was this made for?

Originally built for recalling articles and blog posts you’ve read in the past without remembering the exact title or content. You know you read something relevant, but can’t quite recall the details? revect helps you find it with semantic search! 📚

But we’ve realized it’s incredibly useful for shared and persistent memory for teams and individuals across many AI tools. Whether you’re working solo or collaborating with others, revect becomes your collective knowledge base that any AI can tap into. 🤝

With MCP support, you can use revect as a private way to own your data and recall it seamlessly in any AI system. Your data, your control. 🛡️

🔄 Embedding Model Freedom: Bring any embedding model, any size - from tiny efficient models to massive powerhouses. Swap between OpenAI, local Ollama models, or any OpenAI-compatible provider instantly. revect automatically re-embeds your entire database when you switch, ensuring zero data loss and maximum flexibility! 🪄

  • 🏗️ Single Service Simplicity - Just one container vs complex multi-service alternatives - simpler architecture, easier setup
  • 💾 Portable & Browsable - Store data in SQLite that you can browse anytime with any SQLite viewer
  • 🔍 Semantic Superpowers - Find and retrieve articles from your past with powerful semantic search
  • 💬 AI Memory Bridge - Instantly recall past conversations across different AI providers
  • 🔒 Privacy First - Enjoy complete privacy with fully local, offline operation
  • 🔌 Extensible Ecosystem - Connect with expanding web interfaces and third-party integrations
  • 🤖 Model Agnostic - Use any embedding model or AI provider!

✨ More Reasons to Use

  • 🔄 Lightweight Champion - Minimal dependencies, 100mb complete container size
  • 🤖 Future-Proof Embeddings - Change embedding models anytime, we’ll automatically update all your content
  • 🧩 Plugin Paradise - Extensible architecture with plans for many extensions
  • 🏠 True Data Ownership - Everything stored in a simple SQLite file you control
在 GitHub 查看完整 README

安装

npx mcp-remote http://localhost:8000/mcp
新手?先看 mcpskill.org 教程 →

配置

{ "mcpServers": { "revect": { "command": "npx", "args": ["mcp-remote", "http://localhost:8000/mcp"] } } }