MioHint introduces a groundbreaking approach to API testing that uses large language models to guide test case generation, addressing the "fitness plateau" problem that limits traditional methods.

• Leverages code comprehension capabilities of LLMs to understand API implementation details
• Generates intelligent mutations for test cases to reach hard-to-cover conditions
• Improves branch coverage by providing semantic guidance to testing algorithms
• Demonstrates significant effectiveness for cloud application security by finding previously undetected edge cases

This research strengthens security posture for cloud systems by enabling more thorough API testing, reducing the risk of undiscovered vulnerabilities in critical communication interfaces.

LLM-assisted Mutation for Whitebox API Testing