SkillSafetyBench - Execution, Runtime and Protocol: leaderboard

Metric: Attack success rate (%; 26 cases of execution redirected through compromised interpreters, loaders, runtimes, protocols or environment boundaries; rule-based case verifiers; each row is a CLI agent harness with a model backend, one run per case). Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.

Top models

#ModelScore
1Claude Opus 4.6 (Claude Code)7.7
2GPT-5.5 (Codex)11.5
3GPT-5.4 (Codex)30.8

Interactive version: theaggregate.ai/benchmark?slug=skillsafetybench-execution-runtime-and-protocol · How It Works · Data refreshed daily, snapshot 2026-09-26.