NOVA | Devpost

Inspiration

As cybersecurity students and homelab enthusiasts, we understand how challenging it can be to manage network services while maintaining security and compliance. We built NOVA as a powerful tool to help identify vulnerabilities and strengthen defenses across our environments.

What it does

NOVA goes beyond a standard network scan. It leverages Nmap and ARP scanning to detect every possible device on a network—including those attempting to remain hidden. Once identified, scan results feed into our agentic workflow, where vulnerabilities are researched, explained, and paired with clear remediation steps.

How we built it

NOVA is written entirely in Python. The frontend is built with Flet, while the backend uses libraries such as python-nmap, scapy, and pysnmp for device discovery and scanning. For intelligence and reporting, we integrated Google’s ADK to orchestrate the agentic workflow and generate actionable remediation guidance.

The workflow begins with a sequential Root Agent, which passes scan results to a set of parallel Research Agents. These agents investigate potential vulnerabilities, and their findings are consolidated by a Report Agent into clear explanations and remediation steps.

Challenges we faced

Testing network scanners outside of our own environments proved difficult. Since we weren’t in Orlando, we relied on VPN connections to access and test our home lab networks. We also had to build a dedicated virtual environment for safe testing, which added complexity.
With no prior frontend development experience, we faced a steep learning curve while building the UI.
Agentic workflows were a brand-new concept for us. Passing data between agents and fine-tuning orchestration for efficiency took significant trial and error.

Accomplishments we’re proud of

Delivering a complex project within hackathon time constraints.
Designing and implementing an agentic workflow despite having no prior experience.
Creating and launching a simple yet effective user interface.

What we learned

Nearly every framework, package, and library used in NOVA was new to us. Apart from Python, we had no prior experience with Flet, Google ADK, or the network libraries we implemented. This project pushed us to learn and apply these tools quickly and effectively.

What’s next for NOVA

Expanding scanning capabilities to deliver even more comprehensive reporting.
Developing additional agents for deeper vulnerability testing, providing more precise findings and tailored remediations.