Statically detecting bugs at the binary level has been crucial for the security of Commercial-Off-The-Shelf (COTS) software when source code is not available. However, traditional methods suffer from the inherent limitations of binary translation and static analysis, which hinders their scalability for complex real-world binaries. Recent efforts that leverage large language models (LLMs) for bug detection are still limited by possible hallucination, inaccurate code property retrieval, and insufficient guidance.

In this paper, we propose a new agentic binary analysis framework ClearAgent, which features a novel binary language server that provides both LLM-friendly and analyzer-friendly interfaces to facilitate effective understanding of binary code semantics, enabling effective vulnerability detection. ClearAgent works by automatically interacting with the server and iteratively exploring for buggy locations. For candidate bug reports, ClearAgent further tries to verify the existence of the vulnerability by constructing concrete inputs that can trigger the buggy locations.