Large Language Model-Generated Honeypot Response Dataset for Web Vulnerability Attacks

It mainly includes four typical types of web vulnerability attack requests and their corresponding honeypot response payload data: SQL injection, path traversal, remote code execution, and cross site scripting. The dataset addresses the problems of traditional honeypots relying on fixed response templates, single vulnerability interaction features, and difficulty in adapting to diverse attack requests.

A large language model is introduced to dynamically generate HTTP response data with corresponding vulnerability interaction features based on different vulnerability attack requests, vulnerability types, and contextual information. The data files include four types of data: SQL injection, path traversal, remote code execution, and cross site scripting. Each piece of data contains vulnerability numbers, HTTP response status generated by honeypots, response headers, response bodies, and related generated results. This dataset can be used for research on dynamic honeypots driven by big language models, web vulnerability response simulation, attack trapping, vulnerability interaction feature generation, and network security big model capability evaluation.