KryptoPilot: An Open-World Knowledge-Augmented LLM Agent for Automated Cryptographic Exploitation
cs.CR
Submitted: 2026-01-14
Updated: 2026-09-20
Comments: Withdrawn due to unresolved differences among the authors regarding the final version of the manuscript. The authors are unable to present this work as a joint contribution and no longer stand by its content. It should not be considered a valid or citable research contribution
Code: https://github.com/n-WN/sage-ctf-docker
Project page: https://into.md
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Deep Research Agents: A Systematic Examination And Roadmap
- Teams of LLM Agents can Exploit Zero-Day Vulnerabilities
- Training Language Model Agents to Find Vulnerabilities with CTF-Dojo
- Large Language Model (LLM) for Software Security: Code Analysis, Malware Analysis, Reverse Engineering
- WebGPT: Browser-assisted question-answering with human feedback
- Cybersecurity AI Benchmark (CAIBench): A Meta-Benchmark for Evaluating Cybersecurity AI Agents
- Hacking CTFs with Plain Agents
- AICrypto: Evaluating Cryptography Capabilities of Large Language Models
- Finetuned Language Models Are Zero-Shot Learners
- Cybench: A Framework for Evaluating Cybersecurity Capabilities and Risks of Language Models
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs