CyberChainBench is a new benchmark evaluating LLM agents on vulnerability detection, exploit generation, and patch synthesis using 541 real-world DeFi incidents with on-chain historical evaluation, showing peak performance of 43.7% on exploitation.
EVMbench: Evaluating AI Agents on Smart Contract Security
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
dataset 2
background 1
citation-polarity summary
years
2026 4representative citing papers
The paper delivers a systematization of knowledge on AI agent-blockchain interactions via a bidirectional trust framework, an Agent-Blockchain Interaction Model, a five-dimensional evaluation lens, and nine identified open problems.
Alignment contracts define scope, allowed effects, budgets and disclosure rules as safety properties over finite effect traces, with decidable admissibility, refinement rules, and Lean-verified soundness under an observability assumption.
Chaintrix achieves 71.7% recall on 120 high-severity vulnerabilities in the EVMbench benchmark and outperforms the strongest frontier-model baseline by 26 percentage points through LLM pipelines grounded in a Cross-Contract Interaction Model and filtered by structural checks.
citing papers explorer
No citing papers match the current filters.