NatSec News Scanner

For our Hackathon project NatSec News Scanner, we addressed three critical challenges outlined in the problem set:

• Design a program to scan for and analyze events that could potentially disrupt global supply chains, assessing their impact.
• Develop a tool to monitor online reports of disasters or attacks, verify their authenticity, and predict the affected population sizes.
• Create an API-based, customizable open-source threat intelligence platform that aggregates data from top sources like CrowdStrike, Mandiant, Recorded Future, VirusTotal, and Microsoft to deliver a concise daily summary of the cyber threat landscape.

To tackle these challenges, we developed a multifunctional bot that does all three above. Specifically, this bot integrates three functionalities for comprehensive risk management:

• Supply Chain Disruption Analysis: Identifies and assesses events that could disrupt global supply chains. - Disaster and Attack Monitoring: Monitors reports of disasters or attacks, verifying authenticity and estimating affected populations.
• Cyber Threat Summary: Provides updates on the cyber threat landscape, identifying new threats and vulnerabilities.

Some technical details:

• This bot uses Coze.AI to quickly prototype agents using LLMs. Most of the "code" in this project is a graph composed of nodes on Coze.AI
• After the user enters which domain he or she would like to research on, an LLM triggers the corresponding workflow and returns the result
• All three workflows starts with a number of searches on pre-programmed query keys, and return the top search results (here ~30 entries)
• These search results are then pipelined into an LLM with response to provide a summary

Built With

• coze
• llm
• python