stevemwas/cyber-vulnerabilities

Cyber Vulnerabilities Dataset

This dataset contains a collection of synthetic cybersecurity vulnerability descriptions designed for experimentation with AI systems that analyze and classify security threats.

The dataset was created for the Week 8 AI Engineering project, which demonstrates a multi-agent architecture combining Retrieval Augmented Generation (RAG), LLM reasoning, and ensemble prediction to estimate vulnerability severity.

Dataset Structure

Each record contains the following fields:

• —description – Text description of a cybersecurity vulnerability
• —severity – Severity score between 0 and 10
• —category – Vulnerability category (e.g., Injection, XSS, Memory Corruption)

Example record:

{
  "description": "SQL injection vulnerability in login API allowing attackers to access sensitive database records.",
  "severity": 9.1,
  "category": "Injection"
}