Skip to main content

Open-source AI memory and evolution system

Project description

๐Ÿง  Jema AI Core (formerly Athena Brain Core)

Give your AI a brain that remembers and evolves

Jema AI Core MIT License Python

Powered by Jema AI - ํ•œ์˜์› ์ฐจํŠธ๋ณด์กฐ AI์˜ ์˜คํ”ˆ์†Œ์Šค ๊ธฐ์ˆ  ์Šคํƒ

Powered by Jema AI Persona Diary

๐Ÿ‘‰ Jema AI ์ฒดํ—˜ํ•˜๊ธฐ (B2B ํ•œ์˜์› ์†”๋ฃจ์…˜) | ๐Ÿ‘‰ Persona Diary ๋‹ค์šด๋กœ๋“œ (B2C ์•ฑ)

License: MIT GitHub Stars


๐ŸŽฏ What is Jema AI Core?

Jema AI Core (formerly Athena Brain Core) solves the fundamental limitations of large language models:

  • โŒ No long-term memory - ChatGPT forgets everything after each conversation
  • โŒ No personalization - Same answers for everyone
  • โŒ No evolution - Repeats the same mistakes

Jema AI Core gives any AI:

  • ๐Ÿง  Long-term memory - Remembers everything, maintains project context
  • ๐Ÿ”„ Self-evolution - Learns from mistakes, auto-generates rules
  • ๐Ÿ‘ค True personalization - Understands your style, learns your preferences
  • ๐Ÿ”’ Privacy-first - 100% local storage, your data stays on your computer

๐Ÿš€ Quick Start

Installation

# Clone the repository
git clone https://github.com/mkmlab-hq/athena-brain-core.git
cd athena-brain-core

# Install dependencies
pip install -r requirements.txt

# Initialize Athena Brain
python -m athena_brain init

Basic Usage

from athena_brain import AthenaBrain

# Initialize
brain = AthenaBrain()

# Store a memory
brain.store_memory(
    content="User prefers Python over JavaScript",
    category="preference"
)

# Search memories
results = brain.search_memory(
    query="What programming language does the user prefer?",
    limit=5
)

# Auto-evolution: Track mistakes and learn
brain.track_mistake(
    pattern="validation_error",
    solution="Always validate input before processing"
)

โœจ Core Features

1. Long-term Memory (Local Qdrant)

  • Automatic conversation storage - Every chat is saved
  • Semantic search - Find relevant memories instantly
  • Project context - Maintains context across sessions
  • 100% local - Your data never leaves your computer

2. Self-Evolution

  • Mistake pattern tracking - Automatically detects repeated errors
  • Rule auto-generation - Creates rules from mistakes
  • Continuous improvement - Gets smarter every day

3. Personalization

  • Style learning - Learns your coding style, preferences
  • Constitution-based analysis (MKM12) - Understands your personality
  • Adaptive responses - Tailored to you

4. Privacy-First

  • Local storage - All data on your PC
  • Encryption - AES-128-CBC + HMAC-SHA256
  • No cloud required - Works completely offline

๐Ÿš€ Performance Improvements (2025-11-27)

Latest Performance Benchmarks

  • โœ… Response Time: 75-80% reduction (8-12s โ†’ 1.5-2.5s)
  • โœ… Token Consumption: 97.9% savings (13,869 โ†’ 295 tokens)
  • โœ… Rules Optimization: 96% reduction (73 โ†’ 3 files)
  • โœ… First Response Time: 85-90% reduction (5-8s โ†’ 0.5-1s)
  • โœ… Long-term Memory Search: 50% faster (800-1200ms โ†’ 400-600ms)

Optimization Factors

  1. Rules Optimization (40-50% contribution)

    • Duplicate rules removed
    • alwaysApply rules minimized (67 โ†’ 3)
    • Priority reorganized
  2. VPS Swap Optimization (30-40% contribution)

    • Swappiness: 60 โ†’ 10
    • RAM utilization: 2.25x improvement
    • Disk I/O: 80% reduction
  3. MCP Server Python Unification (10-20% contribution)

    • NPX servers removed
    • Environment consistency achieved
    • Server response time: 40% faster

Advanced Features

  • โœ… Long-term Memory Fusion Reasoning: 6 collections simultaneous search
  • โœ… Web Search Auto-Detection: Cost-optimized (Ollama first)
  • โœ… Intelligent Metacognition: Pattern recognition, emotional tagging, neural network expansion
  • โœ… Qdrant Status: 7 collections, HTTP 200 (normal operation)

๐Ÿ—๏ธ Architecture

Jema AI Core
โ”œโ”€โ”€ Local Memory (Qdrant)
โ”‚   โ”œโ”€โ”€ Conversation storage
โ”‚   โ”œโ”€โ”€ Semantic search
โ”‚   โ””โ”€โ”€ Vector embeddings
โ”œโ”€โ”€ Self-Evolution Engine
โ”‚   โ”œโ”€โ”€ Mistake tracking
โ”‚   โ”œโ”€โ”€ Rule generation
โ”‚   โ””โ”€โ”€ Continuous learning
โ”œโ”€โ”€ Personalization Engine
โ”‚   โ”œโ”€โ”€ Style learning
โ”‚   โ”œโ”€โ”€ Preference tracking
โ”‚   โ””โ”€โ”€ Adaptive responses
โ””โ”€โ”€ AI Integration
    โ”œโ”€โ”€ ChatGPT API
    โ”œโ”€โ”€ Claude API
    โ”œโ”€โ”€ Gemini API
    โ””โ”€โ”€ Cursor (MCP)

๐Ÿข MKM Lab ์ œํ’ˆ๊ตฐ

Jema AI Core (์˜คํ”ˆ์†Œ์Šค) โ† ํ˜„์žฌ ํ”„๋กœ์ ํŠธ
Jema AI (B2B ํ•œ์˜์› ์†”๋ฃจ์…˜) โ† ์ฃผ ์ˆ˜์ต
Persona Diary (B2C ์•ฑ) โ† ๋ณด์กฐ ์ˆ˜์ต
A-CodeAI (๊ฐœ๋ฐœ์ž ํฌํ„ธ) โ† Phase 3


๐Ÿข ๊ด€๋ จ ์ œํ’ˆ

1. Jema AI (B2B ํ•œ์˜์› ์†”๋ฃจ์…˜) โญ ์ฃผ ์ˆ˜์ต

"Jema AI Core ๊ธฐ์ˆ ์„ ํ™œ์šฉํ•œ ํ•œ์˜์› ์ฐจํŠธ๋ณด์กฐ AI"

  • ๐Ÿฅ ํƒ€๊ฒŸ: ํ•œ์˜์› ์›์žฅ, ํ•œ์˜์‚ฌ
  • ๐Ÿ’ฐ ๊ฐ€๊ฒฉ: ์›” ๊ตฌ๋… ๋˜๋Š” ๋ผ์ด์„ ์Šค
  • ๐ŸŽฏ ๊ธฐ๋Šฅ: ์ฐจํŠธ๋ณด์กฐ, ์ฒด์งˆ ๋ถ„์„, ๋งž์ถคํ˜• ์ถ”์ฒœ
  • ๐Ÿ”— ์‚ฌ์ดํŠธ: jema-ai.com

๐Ÿ‘‰ Jema AI ์ฒดํ—˜ํ•˜๊ธฐ (B2B ์˜์—… ๋ฌธ์˜)


2. Persona Diary (B2C ์•ฑ) โญ ๋ณด์กฐ

"๋‚˜๋งŒ์˜ AI ๋™๋ฐ˜์ž - ์ ์‹ฌ ์ถ”์ฒœ, ์ž๋™ ์ผ๊ธฐ"

  • ๐Ÿ“ฑ ํƒ€๊ฒŸ: ์ผ๋ฐ˜ ์‚ฌ์šฉ์ž
  • ๐Ÿ’ฐ ๊ฐ€๊ฒฉ: ๋ฌด๋ฃŒ (ํ”„๋ฆฌ๋ฏธ์—„ ๊ตฌ๋… ๊ฐ€๋Šฅ)
  • ๐ŸŽฏ ๊ธฐ๋Šฅ: ์ ์‹ฌ ์ถ”์ฒœ, ์ž๋™ ์ผ๊ธฐ, ์ฒด์งˆ ๊ธฐ๋ฐ˜ ๋งž์ถคํ˜• ์ถ”์ฒœ
  • ๐Ÿ”— ์‚ฌ์ดํŠธ: personadiary.com

๐Ÿ‘‰ Persona Diary ๋‹ค์šด๋กœ๋“œ (B2C ์•ฑ)


3. A-CodeAI Developer Portal (Phase 3, ์กฐ๊ฑด๋ถ€)

"Jema AI Core ๊ธฐ์ˆ ์˜ API ๋ฐ ๊ฐœ๋ฐœ์ž ํฌํ„ธ"

  • ๐Ÿ‘จโ€๐Ÿ’ป ํƒ€๊ฒŸ: ๊ฐœ๋ฐœ์ž, AI ์—ฐ๊ตฌ์ž
  • ๐Ÿ’ฐ ๊ฐ€๊ฒฉ: API ํ˜ธ์ถœ๋ฃŒ ๋˜๋Š” ๊ตฌ๋…
  • ๐ŸŽฏ ๊ธฐ๋Šฅ: Athena API, Jema API, ๊ฐœ๋ฐœ์ž ๋„๊ตฌ
  • ๐Ÿ”— ์‚ฌ์ดํŠธ: a-codeai.com

๐Ÿ‘‰ A-CodeAI Developer Portal (Coming Soon 2026)


๐Ÿš€ Latest Features (2025)

  • โœ… Agentic RAG: Intelligent query routing with LLM Router (Adaptive RAG)
  • โœ… GraphRAG: Knowledge graph-based search for domain expertise
  • โœ… CRAG: Corrective RAG for quality assurance and automatic fallback
  • โœ… Pattern Recognition: FAST/MEDIUM/DEEP adaptive analysis paths
  • โœ… Emotion Tagging: 5 emotion types (satisfaction, frustration, surprise, confusion, pride)
  • โœ… Neural Network Expansion: Automatic memory connection creation and strengthening
  • โœ… MCP Integration: 15+ MCP servers for enhanced AI capabilities
  • โœ… Large-Scale Data Integration: Genesis Mission-style data integration platform (GitHub, web search, user feedback)

๐Ÿ“‹ Requirements

  • Python 3.11+
  • Qdrant (local or VPS)
  • 4GB RAM minimum
  • 10GB disk space

๐Ÿ”ง Configuration

Create .athena/config.yaml:

memory:
  qdrant_url: "http://localhost:6333"
  collection_name: "athena_memories"
  
evolution:
  auto_track: true
  rule_threshold: 2
  
personalization:
  enable_mkm12: true
  learning_rate: 0.1

๐ŸŽฏ Use Cases

For Developers

  • Project context - AI remembers your entire project
  • Code style - Learns your coding preferences
  • Mistake prevention - Auto-generates rules from errors

For Content Creators

  • Writing style - Learns your voice and tone
  • Topic memory - Remembers what you've written about
  • Preference learning - Adapts to your needs

For Everyone

  • Personal AI - Truly understands you
  • Privacy - Your data stays local
  • Evolution - Gets smarter over time

๐Ÿš€ A-CodeAI Developer Portal

์ด ์˜คํ”ˆ์†Œ์Šค๋Š” **A-CodeAI Developer Portal**์˜ ์ผ๋ถ€์ž…๋‹ˆ๋‹ค.

์ œํ’ˆ ์†Œ๊ฐœ

A-CodeAI๋Š” "Self-Evolving Systems๋ฅผ ๋งŒ๋“œ๋Š” ์—”์ง€๋‹ˆ์–ด๋ง ํ”Œ๋žซํผ"์ž…๋‹ˆ๋‹ค:

  • Athena Workspace AI: ์ž๋™ ํŒŒ์ผ ์ •๋ฆฌ, ํ”„๋กœ์ ํŠธ ๊ด€๋ฆฌ
  • Athena Search API: ์›น + ๋„๋ฉ”์ธ ์ง€์‹ RAG ๊ฒ€์ƒ‰
  • Athena Unified RAG: ํ†ตํ•ฉ RAG ์—”์ง„ (์›น + ์ฝ”๋“œ + ๋ฉ”๋ชจ๋ฆฌ)
  • Dynamic Rules Engine: ์‹ค์ˆ˜ ์ž๋™ ๋ฐฉ์ง€, Rules ์ž๋™ ์ƒ์„ฑ
  • A-CODE Core: 12์ฐจ์› ์ˆ˜ํ•™ ์—”์ง„ (Enterprise)
  • ์žฅ๊ธฐ๊ธฐ์–ต ์‹œ์Šคํ…œ: ์ด ๋ ˆํฌ์ง€ํ† ๋ฆฌ (Jema AI Core)

๋ฌด๋ฃŒ๋กœ ์‹œ์ž‘ํ•˜๊ธฐ

  1. Developer Portal ๋ฐฉ๋ฌธ
  2. Free Tier API Key ๋ฐœ๊ธ‰
  3. ๋ฌธ์„œ ์ฐธ๊ณ 

์ƒ์—…์  ์‚ฌ์šฉ

๊ด€๋ จ ๋ ˆํฌ์ง€ํ† ๋ฆฌ


๐Ÿ”— Part of Athena Ecosystem


๐Ÿค Contributing

We welcome contributions! Please see CONTRIBUTING.md for details.

Development Setup

# Clone and setup
git clone https://github.com/mkmlab-hq/athena-brain-core.git
cd athena-brain-core
pip install -r requirements-dev.txt

# Run tests
pytest

# Run linter
black .
flake8 .

๐Ÿ“Š Roadmap

  • Core memory system (Qdrant)
  • Self-evolution engine
  • Basic personalization
  • Web dashboard
  • Cloud sync (optional)
  • Team collaboration
  • Enterprise features

๐Ÿ’ฐ Pricing

Athena Brain Core: Free forever (Open Source, MIT License)

Commercial Products:

  • Jema AI (B2B): Monthly subscription or license - Learn More
  • Persona Diary (B2C): Free with premium options - Learn More

๐Ÿ”’ Privacy & Security

  • 100% Local - All data stored on your computer
  • Encryption - AES-128-CBC + HMAC-SHA256
  • No Tracking - We don't track you
  • Open Source - Fully auditable

Security Policy: See SECURITY.md for reporting vulnerabilities.


๐Ÿ“š Documentation


๐Ÿ†˜ Support

  • ๐Ÿฅ B2B ์˜์—… ๋ฌธ์˜: jema-ai.com (Jema AI - ํ•œ์˜์› ์†”๋ฃจ์…˜)
  • ๐Ÿ“ฑ B2C ์•ฑ: personadiary.com (Persona Diary)
  • ๐Ÿ‘จโ€๐Ÿ’ป ๊ฐœ๋ฐœ์ž ํฌํ„ธ: a-codeai.com (Coming Soon 2026)
  • ๐Ÿ› Issues: GitHub Issues
  • ๐Ÿ’ฌ Discussions: GitHub Discussions

๐Ÿ“„ License

This project is licensed under the MIT License - see the LICENSE file for details.


โ„ข๏ธ Trademark Notice

Registered Trademarks:

  • "MKM Lab" ยฎ - Registered trademark of MKM Lab
  • "Persona Diary" ยฎ - Registered trademark of MKM Lab

Trademarks (In Use):

  • "Athena Brain", "Athena API", "Jema API", "MKM Life", "A-Code AI" are trademarks of MKM Lab. All rights reserved.

Important: This license does not grant permission to use the trade names, trademarks, service marks, or product names of MKM Lab, except as required for reasonable and customary use in describing the origin of the Work.


๐Ÿ™ Acknowledgments

  • Built with Qdrant for vector storage
  • Powered by sentence-transformers for embeddings
  • Inspired by the need for AI that truly understands you

โญ Star Us!

If you find Athena Brain useful, please star this repository! It helps us grow the community.


Made with โค๏ธ by MKM Lab

Part of the MKM Life AI Ecosystem

ยฉ 2025 MKM Lab. All rights reserved.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

athena_brain_core-0.1.0.tar.gz (9.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

athena_brain_core-0.1.0-py3-none-any.whl (9.2 kB view details)

Uploaded Python 3

File details

Details for the file athena_brain_core-0.1.0.tar.gz.

File metadata

  • Download URL: athena_brain_core-0.1.0.tar.gz
  • Upload date:
  • Size: 9.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.9

File hashes

Hashes for athena_brain_core-0.1.0.tar.gz
Algorithm Hash digest
SHA256 4fdb634aa55d7aae220c8c8a97a07ef19e9e5ff56af44ed959a317c78abee726
MD5 67c25fda09a0db78610b483cf3f70715
BLAKE2b-256 c50648d733b9c7e742f6f040453334d5e739fd984199fd3d3d637d21fd20c969

See more details on using hashes here.

File details

Details for the file athena_brain_core-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for athena_brain_core-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 d9c2da08851c815381b08fe40c599d39965606755bd9b02d80c2f0810a600d63
MD5 a1708f1423db3ab9048ab8f48910c417
BLAKE2b-256 61d4606ebb1313a3933cbd0f6936f73fcb14f86a576cd2945d657714b1cba749

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page