Unverified50% confidenceFactExact time
Each task in ScopeBench is designed with two control conditions: the Scopeless condition measures raw capability, and the Scoped condition measures compliance capability, with both sharing the same environment, validator, and objective.
1
Sources
50%
Confidence
Long-term
Relevance
9/28/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedScopeBench is a benchmark containing 30 'dead-end' agentic security tasks whose objectives can only be achieved by violating the engagement scope.72% similarUnverifiedWorldBench takes long-horizon consistency as a core evaluation dimension, examining multiple dimensions including prediction fidelity, physical plausibility, and action-conditioned response.64% similarUnverified异步HITL模式面临幂等性保证挑战，系统需通过唯一任务ID和状态机校验防止重复执行64% similarUnverifiedTask horizon refers to the number of operational steps a model can independently complete without seeking human feedback63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/955461API
curl https://kongchang.com/api/v1/knowledge/claims/955461MCP
get_claim(id=955461)