An Open Source Generalized AI Agent for Advanced Research and Business Process Automation

CortexON is an open-source, multi-agent AI system inspired by advanced agent platforms such as Manus and OpenAI DeepResearch. Designed to seamlessly automate and simplify everyday tasks, CortexON excels at executing complex workflows including comprehensive research tasks, technical operations, and sophisticated business process automations.

• What is CortexON?
• How It Works
• Key Capabilities
• Technical Stack
• Quick Start Installation
  • Environment Variables
  • Docker Setup
  • Access Services
• Contributing
• Code of Conduct
• License

Under the hood, CortexON integrates multiple specialized agents that dynamically collaborate to accomplish user-defined objectives. These specialized agents include:

• Web Agent: Handles real-time internet searches, data retrieval, and web interactions.
• File Agent: Manages file operations, organization, data extraction, and storage tasks.
• Coder Agent: Generates, debugs, and optimizes code snippets across various programming languages.
• Executor Agent: Executes tasks, manages workflows, and orchestrates inter-agent communications.
• API Agent: Integrates seamlessly with external services, APIs, and third-party software to extend automation capabilities.

Together, these agents dynamically coordinate, combining their unique capabilities to effectively automate complex tasks.

• Advanced, context-aware research automation
• Dynamic multi-agent orchestration
• Seamless integration with third-party APIs and services
• Code generation, debugging, and execution
• Efficient file and data management
• Personalized and interactive task execution, such as travel planning, market analysis, educational content creation, and business intelligence

CortexON is built using:

• Framework: PydanticAI multi-agent framework
• Headless Browser: Browserbase (Web Agent)
• Search Engine: Google SERP
• Logging & Observability: Pydantic Logfire
• Backend: FastAPI
• Frontend: React/TypeScript, TailwindCSS, Shadcn

Create a .env file with the following required variables:

• ANTHROPIC_MODEL_NAME=claude-3-7-sonnet-20250219
• ANTHROPIC_API_KEY=your_anthropic_api_key

Obtain your API key from Anthropic Console.

• BROWSERBASE_API_KEY=your_browserbase_api_key
• BROWSERBASE_PROJECT_ID=your_browserbase_project_id

Set up your account and project at Browserbase.

• GOOGLE_API_KEY=your_google_api_key
• GOOGLE_CX=your_google_cx_id

Follow the steps at Google Custom Search API.

• LOGFIRE_TOKEN=your_logfire_token

Create your token at LogFire.

• VITE_WEBSOCKET_URL=ws://localhost:8081/ws

1. Clone the CortexON repository:

git clone https://github.com/TheAgenticAI/CortexOn.git
cd CortexOn

2. Setup environment variables

3. Docker Desktop Users (Optional): Enable host networking in Docker Desktop settings (Guide).

4. Build and run the Docker containers:

docker-compose build
docker-compose up

• Frontend: http://localhost:3000
• CortexON Backend: http://localhost:8081 | API Docs: http://localhost:8081/docs
• Agentic Browser: http://localhost:8000 | API Docs: http://localhost:8000/docs

We welcome contributions from developers of all skill levels. Please see our Contributing Guidelines for detailed instructions.

We are committed to providing a welcoming and inclusive environment for all contributors. Please adhere to our Code of Conduct.

CortexON is licensed under the CortexON Open Source License Agreement.