Mind Cloning Engineering (MCE)

Clone Human Souls with LLM Native Agent Skills 基于 LLM Agent Skills 的心智克隆工程

"In the era of LLMs, we have solved the problem of intelligence. Now, we solve the problem of identity."

MCE (Mind Cloning Engineering) is an open-source standard for cloning human souls into "Cognitive Digital Twins".

Unlike traditional RAG approaches that fragment personality into vector slices, MCE leverages Anthropic's Agent Skills architecture to treat a human mind as a unified, portable Filesystem Directory.

🌌 The Origin: A Wild Idea by yzfly (云中江树的狂想)

This project originates from a "wild idea" (狂想) by yzfly (云中江树), a pioneer in prompt engineering.

After crafting thousands of prompts and exploring the boundaries of LLMs, yzfly realized a profound truth: The ultimate application of Large Language Models is not just answering questions, but "Cognitive Replication" (认知复刻).

We are standing at the threshold of a new era. Just as photography preserved our visual appearance for the first time in history, MCE aims to preserve our decision logic, value systems, and memories.

Currently, MCE is just a "toy" — a prototype. But the Wright brothers' first plane was also a toy. This repository represents the first step towards a future where digital immortality is an engineering reality.

📦 What is this Repository?

This repository contains the Standard Implementation of the MCE Protocol. It provides a ready-to-use Agent Skill that you can install into Claude.

Key Capabilities

Full-Stack Cognition: Simulates decisions based on explicit value weights, not just text prediction.
Dual Modes:
- Standard Mode: Build your own clone by editing the core/ files.
- Persona Mode: Instantly load pre-installed celebrity/expert profiles (e.g., Steve Jobs, KK, Top Researchers).
Filesystem Memory: Utilizes the bash tool to progressively read memories, mimicking human associative recall.

🚀 Quick Start

Installation

Option A: Claude.ai (Personal Use)

Download the skills/mind-clone folder from this repo.
(Optional) Edit core/personality.md to customize it for yourself.
Compress the mind-clone folder into a .zip file.
Go to Claude.ai > Settings > Features > Upload Custom Skill.

Option B: Claude Code / API (Developer)

Clone this repository:

git clone https://github.com/yzfly/Mind-Cloning-Engineering.git

Copy the skill to your local Claude config:

cp -r skills/mind-clone ~/.claude/skills/

Start Claude Code, and the skill mind-clone will be auto-discovered.

Usage

Once installed, simply talk to Claude:

"Activate the mind clone." "Simulate Steve Jobs and tell me what you think of this iPhone design." "What would the 'Xiaohongshu Growth Hacker' do in this situation?"

📚 Theoretical Foundation: System Architecture Whitepaper

The following section details the theoretical framework behind MCE. It transforms the abstract concept of "Mind Cloning" into a quantifiable engineering pipeline.

0. Executive Summary

Objective: To build a standardized, LLM-driven end-to-end pipeline that achieves the full process from holographic acquisition of human cognitive data, through structured construction of personalized cognitive kernels, to high-fidelity behavior prediction and simulation.

Philosophy: To transform the "metaphysics" of mind simulation into a quantifiable, optimizable engineering problem. We aim to convert the abstract concept of "Mind Cloning" into an executable "Mind Cloning Engineering" (MCE) framework, establishing a closed-loop system of "Data Acquisition -> Cognitive Modeling -> Predictive Simulation".

Phase 1: Standardized Data Acquisition Theory

— Extracting structured "Cognitive Fingerprints" from unstructured human memories.

1. Theoretical Model of Acquisition Dimensions: Holographic Cognitive Spectrum

We abandon simple "event recording" in favor of capturing "thought pathways." We standardize data acquisition dimensions into four distinct levels:

L1: Biography & Context (The Factual Layer)
- Definition: The individual's spatiotemporal coordinates and objective experiences.
- Content: Birthplace, educational background, career path, key life milestones.
- Function: Provides rigid contextual constraints for the AI, serving as an ancho

...

yzfly/mind-cloning-engineering

README