Security Copilot

• 

• 

• 

• 

  Agent response

Inspiration

Phishing attacks are becoming more sophisticated, while traditional security tools often rely on static URL reputation or predefined rules. We wanted to build something more intelligent: an AI agent that can actually investigate a suspicious website like a security analyst instead of simply classifying its URL.

What it does

Security Copilot is an agentic AI-powered Chrome extension that autonomously investigates websites for phishing and other web threats. It analyzes factors such as URL and domain information, redirects, DOM structure, forms, page content, screenshots, WHOIS data, and VirusTotal intelligence to identify suspicious behavior and provide an explainable risk assessment.

How we built it

We built Security Copilot as a Chrome extension connected to an AI-powered backend. The agent coordinates multiple investigation steps, collects security signals from different sources, analyzes webpage behavior and structure, and uses AI reasoning to correlate the evidence before generating a final verdict and explanation.

Challenges we ran into

The biggest challenge was making the system genuinely agentic rather than just creating a fixed sequence of security checks. We also had to handle untrusted webpages, dynamically changing websites, redirects, incomplete threat-intelligence data, and the challenge of combining many different signals into one reliable assessment.

Accomplishments that we're proud of

We successfully built a working end-to-end prototype capable of autonomously investigating suspicious websites. Instead of simply saying "phishing" or "safe," Security Copilot provides the evidence behind its decision, making the detection process more transparent and useful to users.

What we learned

We learned how to combine agentic AI, browser automation, cybersecurity intelligence, web analysis, and threat detection into a single system. We also learned that reliable security decisions require correlating multiple independent signals rather than depending on a single AI prediction or reputation database.

What's next for Security Copilot

We plan to improve detection accuracy, expand anomaly detection capabilities, add more threat-intelligence sources, strengthen sandbox isolation, and build a larger dataset for evaluation. Eventually, we want Security Copilot to evolve from a phishing detector into a fully autonomous browser security analyst capable of investigating a wider range of web-based threats.

Built With

• agentic
• agents
• ai
• analysis
• anomaly
• automation
• browser
• chrome
• css
• cybersecurity
• detection
• dom
• extension
• fastapi
• generative
• html
• intelligence
• javascript
• langgraph
• llm
• phishing
• python
• threat
• virustotal
• whois