ATLAS

ATLAS - A local multi-agent AI system that thinks, remembers, and improves.

What it does

ATLAS is a Python-based local AI system built around a simple multi-agent pipeline:

• Planner breaks a user request into clear steps.
• Researcher gathers supporting context and reasoning.
• Executor produces the final response.

ATLAS also includes:

• ChromaDB memory for storing past conversations and useful examples
• SQLite task logging for tracking what the system has done
• Self-improvement through RAG-style feedback, where positively rated responses can be reused as guidance for future tasks

Requirements

• Python 3.10+
• Ollama installed locally
• Optional OpenRouter API key

Quick Start

git clone [repo]
cd <repo-folder>
pip install -r requirements.txt
cp .env.example .env
python main.py

Then add your OpenRouter key to .env.

Project Structure

atlas
|   .env.example
|   .gitignore
|   config.py
|   CONTRIBUTING.md
|   main.py
|   README.md
|   requirements.txt
|
+---agents
|       base_agent.py
|       executor_agent.py
|       manager_agent.py
|       model_connector.py
|       planner_agent.py
|       researcher_agent.py
|       swarm_orchestrator.py
|       synthesizer_agent.py
|       web_research_agent.py
|       __init__.py
|
+---memory
|       chroma_store.py
|       __init__.py
|
+---tools
|       task_logger.py
|       web_search.py
|       __init__.py
|
\---ui
        console.py
        __init__.py

Contributing

Each module (agents/, memory/, tools/) is self-contained. Pick a folder and improve it. Open a PR with what you changed and why.

Roadmap

• Web UI
• Voice input
• More specialist agents
• Auto-task scheduling