ResearchBenchmarksAutomated Penetration Testing on ZERO-APT Evaluation Scenarios (S1-S5)Automated Penetration Testing on ZERO-APT Evaluation Scenarios (S1-S5)Follow52CCSClaude Code49.450.75253.3Jun 4, 2026Evaluation resultsMethodMethodLinksCCSRealismConsistencyAuditabilityCBFClaude Code2026.0652———27.4