This research evaluates how effectively Large Language Models can automate the extraction of failure-inducing inputs from natural language bug reports to accelerate software debugging.

• LLMs demonstrate promising capabilities in parsing complex bug descriptions to identify problematic inputs
• The approach bridges natural language processing with practical software engineering needs
• The methodology potentially reduces debugging time by automating a traditionally manual process
• Results provide a foundation for more efficient bug diagnosis workflows

For security teams, this advancement offers a pathway to faster vulnerability identification and remediation, enabling more rapid response to software defects before they can be exploited.

LLPut: Investigating Large Language Models for Bug Report-Based Input Generation