Reproduce “Thinking in Graphs: Neo4j-Backed Exploit-Chain Reasoning for LLM Penetration-Testing Agents” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Joas Antonio dos Santos Barbosa

ChainGraph stores findings and relationships in a persistent graph, letting agents query routes to a goal. The authors connect unmodified coding agents through MCP to Neo4j and keep evidence in files. In a synthetic, hand-authored Active Directory case, a query returned an eight-node, seven-hop path from a Medium finding to forest root; autonomous discovery and execution were not tested.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.