Health-Yang

Health-Yang / MineEcho

Public

Local-first Memory OS for personal AI assistants with L0-L3 memory, Wiki++ knowledge, skill routing, and TokenLess context compression.

ai-agent ai-assistant knowledge-graph local-first memory
15
2
89% credibility
Found May 29, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

MineEcho is a personal AI assistant framework that runs entirely on your own computer. Unlike chatbots that start fresh every time, it builds a persistent memory of your conversations and learnings. It can import and organize your documents, files, and notes into a searchable knowledge base. You can connect it to various AI services (like MiniMax, DeepSeek, or similar) by providing your own account, and it stays completely local—your data never leaves your machine. The project includes a desktop application, a web-based console, and supports adding "skills" that extend what your assistant can do.

How It Works

1
💡 Discover MineEcho through word of mouth

A friend tells you about an AI assistant that actually remembers things about you, unlike other chatbots that forget everything.

2
📦 Download and install on your computer

You grab the installer for your computer (Windows, Mac, or Linux), and it sets itself up quietly on your machine—no cloud account needed.

3
🎨 Open the console and configure your AI brain

A friendly window appears where you type in the name of your favorite AI service, like giving your assistant a name and personality.

4
🧠 Your assistant starts building a memory

As you chat, it quietly saves what you talk about—your projects, preferences, the things you keep asking about—so it learns your world.

5
📚 Drop in your documents and knowledge

You drag in your work files, project notes, or any documents you want your assistant to know about. It reads through them and stores the important parts.

6
Choose how you want to work
💬
Ask questions and get help

Just chat naturally and watch your assistant pull from your memory and documents to give smarter answers.

🛠️
Add specialized skills

Browse ready-made skill packages that give your assistant superpowers for specific tasks like coding or writing.

✨ Everything works together, privately

Your AI assistant remembers your history, knows your documents, and stays completely on your own computer—no one else can see it.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 15 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is MineEcho?

MineEcho is a local-first Memory OS for personal AI assistants built in TypeScript. It solves the "forgetful chat UI" problem by maintaining persistent memory across sessions through a layered L0-L3 memory tree, importing knowledge into an inspectable Wiki++ system, and routing user requests to skills rather than dumping everything into one general-purpose prompt. The project also includes TokenLess, a compression layer that reduces token costs by intelligently trimming long tool outputs and document extractions before they enter context. You run it as a local BFF service with a web console, or package it as a desktop app via Electron.

Why is it gaining traction?

The hook is the memory architecture. Most AI assistants treat every conversation as disposable, but MineEcho separates raw interaction memory from distilled summaries, so your assistant actually remembers what you told it last week. Combined with a knowledge base that uses four retrieval channels instead of simple chunk embedding, and a cost control layer that keeps errors and actionable lines while discarding repetition, this targets developers who want persistent, cost-efficient personal AI workflows without sending everything to the cloud.

Who should use this?

Developers building personal AI assistants who want memory that survives beyond single sessions. Knowledge workers managing private documentation who need retrieval that goes beyond basic RAG. Teams evaluating local-first AI infrastructure who want to experiment before committing to cloud-based alternatives. Not suitable for production enterprise deployments yet given the early star count.

Verdict

MineEcho addresses a real gap in the personal AI assistant space with a thoughtful multi-layered memory architecture and genuine local-first design. The documentation is thorough and the feature set is coherent. However, with 15 stars and a credibility score of 0.9%, this is early-stage software. The PolyForm Noncommercial License also restricts commercial use, which limits adoption scope. Worth exploring as a technical proof-of-concept for local-first AI memory systems, but wait for more community validation before betting production workflows on it.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.