Unverified50% confidenceFactExact time
The benchmark consists of 200 manually constructed task items in four formats—route, classify, judge, and ground—of which one-fifth of the questions can only be answered correctly with ESCALATE.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe researchers evaluated on a synthetic task set constructed by referencing the published structure of TriggerBench, which contains 48 blind-written dialogues and 175 tasks.71% similarUnverifiedA verifiable prompt should clearly specify four things: what the task is, which source material to refer to, what format the output should be in, and how to handle missing information.65% similarUnverified该项目单个仓库内置278个技能、67个子智能体、94条现成命令，覆盖12种语言生态65% similarUnverifiedThe optimal strategy for discriminative tasks consists of a set of thresholds, one corresponding to each class.65% similarUnverifiedWhen building a tool site, one should first plan more than 100 small tools that the target audience genuinely needs and maintain a focused theme.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002959API
curl https://kongchang.com/api/v1/knowledge/claims/1002959MCP
get_claim(id=1002959)