A reward-driven, PPO-finetuned LLM pipeline claims to generate diverse, evasive webshell payloads with higher escape rates than prompt-engineering baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
A Reward-driven Automated Webshell Malicious-code Generator for Red-teaming
A reward-driven, PPO-finetuned LLM pipeline claims to generate diverse, evasive webshell payloads with higher escape rates than prompt-engineering baselines.