GONDAR finds four times more Java vulnerabilities than Jazzer by identifying reachable sinks via CWE scanning and LLM filtering then using collaborative exploration and exploitation agents with a coverage-guided fuzzer.
Harnessing large language models for seed generation in greybox fuzzing
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CR 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
LLM-assisted mutation testing reproduced 10 of 13 known PKCS#1 v1.5 violation categories and one new discrepancy but was limited by 82.5% hallucination rate in generated scripts.
citing papers explorer
-
Contextualizing Sink Knowledge for Java Vulnerability Discovery
GONDAR finds four times more Java vulnerabilities than Jazzer by identifying reachable sinks via CWE scanning and LLM filtering then using collaborative exploration and exploitation agents with a coverage-guided fuzzer.
-
Evaluating the Effectiveness of LLMs in Aiding Compliance Testing of PKCS#1-v1.5
LLM-assisted mutation testing reproduced 10 of 13 known PKCS#1 v1.5 violation categories and one new discrepancy but was limited by 82.5% hallucination rate in generated scripts.