Chimera Protocol �

CodeCraze Hackathon 2024 Submission

A neural-themed AI memory management system that unifies multiple LLM providers into one collaborative platform with 3D brain visualization and persistent memory injection.

Watch the Demo Video

🎯 The Problem We Solved

The AI industry is fragmented. OpenAI, Anthropic, Google, and others all operate in isolation: different APIs, different strengths, different ecosystems. Users are forced to choose one provider or juggle multiple accounts and interfaces. Worse, every time you switch models, you lose all context and have to start from scratch.

Chimera Protocol unifies them all.

💡 What Chimera Protocol Does

The Problem	Our Solution
AI providers don't talk to each other	Unified Multi-LLM Router: One interface for all providers
Each model has different strengths	Model Switching: Use GPT-4 for code, Claude for writing, Gemini for analysis
AI has no persistent memory	Memory Injection System: Store knowledge and inject it into any model
Teams can't share AI context	Collaborative Workspaces: Teams share memories, conversations, and API keys
API keys scattered everywhere	Encrypted Key Vault: One secure place for all provider credentials

🧬 The Chimera Architecture

        🧠 OpenAI (GPT-4o, GPT-4)
       /
      🧠 Anthropic (Claude 3.5 Sonnet)
     /
🦁--🧠 Google (Gemini 2.0 Flash)
     \
      🧠 Groq (Llama 3.3 70B) - FREE
       \
        🧠 DeepSeek (DeepSeek Chat)

Each "head" is a different AI provider, but they all share:

• The same memory bank: Inject context into any model
• The same workspace: Team members collaborate across providers
• The same conversation history: Switch models mid-conversation
• The same secure storage: One encrypted vault for all API keys

✨ Key Features

Multi-LLM Support

Provider	Models	Status
OpenAI	GPT-4o, GPT-4, GPT-3.5-turbo	✅
Anthropic	Claude 3.5 Sonnet, Claude 3	✅
Google	Gemini 2.0 Flash, Gemini Pro	✅
Groq	Llama 3.3 70B, Mixtral 8x7B	✅ FREE
DeepSeek	DeepSeek Chat, DeepSeek Coder	✅

Memory System

• Manual Creation: Create memories with titles, content, and tags
• URL Import: Scrape and store content from web pages (Basic & Advanced modes)
• File Import: Import from PDF, DOCX, TXT, and more
• Memory Injection: Inject relevant context into any AI conversation
• Semantic Search: Find relevant memories by content

3D Brain Visualization

• Interactive Three.js brain model
• AI models displayed as glowing neural nodes
• Real-time connection visualization
• Cyberpunk-inspired interface

Workspace & Team Collaboration

• Isolated workspaces per project
• Team collaboration with invites
• Role-based access control (Admin, Researcher, Observer)
• Real-time presence indicators

📦 Repository Structure

Chimera_Protocol/
├── Backend/     # Django REST API
└── Frontend/    # React/TypeScript UI

Component	Tech Stack	Description
Backend	Django 5, Python 3.11+, PostgreSQL	REST API, LLM routing, memory storage
Frontend	React 18, TypeScript 5, Three.js, Tailwind	3D brain visualization, chat interface

🚀 Quick Start

Clone with Submodules

git clone --recurse-submodules https://github.com/ahammadshawki8/Chimera_Protocol.git
cd Chimera_Protocol

Backend Setup

cd Backend
python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env
python generate_key.py
python manage.py migrate
python manage.py runserver

Frontend Setup

cd Frontend
npm install
cp .env.example .env.local
npm run dev

🔧 Environment Variables

Backend (.env)

SECRET_KEY=your-django-secret-key
DEBUG=True
DATABASE_URL=postgres://user:pass@localhost:5432/chimera
ENCRYPTION_KEY=your-fernet-key

Frontend (.env.local)

VITE_API_BASE_URL=http://127.0.0.1:8000/api

🔌 API Highlights

Endpoint	Description
POST /api/auth/register	User registration
POST /api/auth/login	JWT authentication
GET /api/workspaces	List workspaces
POST /api/conversations/{id}/messages	Send chat message
POST /api/conversations/{id}/inject-memory	Inject memory context
POST /api/memories/search	Search memories
GET /api/models/available	List connected AI models

🔐 Security

• JWT authentication with refresh tokens
• Fernet encryption for API keys at rest
• CORS configuration
• Input validation and sanitization
• SQL injection prevention via Django ORM

🎨 Screenshots

The interface features a cyberpunk-inspired design with:

• Neon green accents on dark backgrounds
• Angular frames and glowing borders
• Interactive 3D brain for model selection
• Responsive design for all screen sizes

📄 License

MIT License - see LICENSE file

👥 Team

Built for CodeCraze Hackathon by:

• ahammadshawki8
• mumtio

---

Chimera Protocol: Unifying AI, one model at a time. 🧠⚡