Root-Cause-Driven Vulnerability Repair
Status: First-author paper, under submission. Ph.D. research at ASU SEFCOM.
- Designed a two-stage repair pipeline combining static and dynamic analysis to locate potential root causes, capturing defects that span multiple code locations and indirect calls missed by static-only tools such as CodeQL.
- Supplied root-cause candidates to a lightweight patching agent with repository-search and file-editing tools, so repairs are informed by the relevant code context.
- Built a manually annotated patch-quality dataset covering 178 C/C++ vulnerabilities to compare repairs generated by multiple agents.
- Improved patch quality by 20% over baseline methods, validated by majority vote of three human experts.
Related: LLM-assisted fuzzing for occluded bugs
An LLM-assisted patching + AFL++ pipeline that surfaces vulnerabilities masked by more easily triggered crashes. Automated crash clustering, root-cause hints, LLM-guided edits, and re-fuzzing reduced false positives from 90% to 30% and found 20+ distinct occluded vulnerabilities per target in Objdump, FFmpeg, and GPAC.