allumia
v0.2.13
Published
[](https://github.com/ruben-angelo/allumIA/actions/workflows/ci.yml) [](https://opensource.org/licenses/MI
Downloads
2,583
Readme
allumIA
A privacy-focused, local-first AI interface for Ollama. Use it as a beautiful web dashboard or directly in your terminal. AllumIA features RAG (Retrieval Augmented Generation), Web Search, and a long-term memory system — all running 100% on your machine.
🌟 Features
- Dual Interaction Modes: Switch between a sleek Web UI and a powerful CLI.
- Privacy First: No cloud dependencies. Your data, documents, and secrets never leave your hardware.
- Intelligent RAG: Chat with your local documents using high-performance vector storage (LanceDB).
- Local Web Search: Deep research with Serper or self-hosted SearXNG.
- Smart Memory: Remembers context from previous conversations.
- MCP Compatible: Extensible through the Model Context Protocol.
🚀 Quick Start
Prerequisites
- Node.js 20.x or higher.
- Ollama (Default:
http://localhost:11434). - Models:
- Chat:
ollama pull llama3(or your preferred model). - RAG:
ollama pull nomic-embed-text.
- Chat:
Installation (Global)
npm install -g allumiaThe installation automatically performs a silent initialization (configures ~/.config/allumia/).
🛠 Usage
Web Dashboard
Start the local server and open your browser:
# Run in foreground
allumia serve
# Run in background and open browser
allumia serve bg --open
# Stop background server
allumia serve stopCLI Chat (Terminal Only)
Chat directly without starting the web UI:
allumia --model llama3Configuration
List models and save your preference:
allumia model🛡 Security
AllumIA is designed for local environments.
- Local-first Authentication: Requests to localhost are protected by a
SameSite=StrictandHttpOnlycookie. - Secret Isolation: Your
ALLUMIA_SECRET_KEYis never exposed to the frontend JavaScript, mitigating XSS risks. - Resource Limiting: Built-in protection against common local-first vectors like CSRF on local host.
For more details, see docs/shared/SECURITY.md.
🏗 Architecture
For a detailed breakdown of the internal layers (UI, Engine, Tools, Runtime), please refer to docs/shared/ARCHITECTURE.md.
🤝 Contributing
- Clone the repo.
- Install dependencies:
npm install. - Start development server:
npm run dev. - Run tests:
npm test.
📄 License
This project is licensed under the MIT License - see the LICENSE file for details.
