Unverified50% confidenceTradeoffExact time
模型自主触发Skills时，全量注入所有Skill描述的方案在Skills数量超过数十个后性能显著下降
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
Unverified在工具数量超过20个的场景下，工具描述的语义歧义度与模型选择准确率之间存在显著的负相关关系68% similarUnverified研究表明当ReAct循环超过10步时，模型的任务完成率会出现显著下降64% similarUnverified消融实验显示，若继续训练主干通用知识掉10%、数学掉17.8%；若两塔共享权重则通用任务掉26%，其余指标全线崩溃64% similarUnverifiedWhen the number of MCP tools is excessive, AI reliability drops from 100% to 72% because too many tool descriptions are injected into the context.64% similarUnverifiedAnthropic's research shows that even mild alignment training can cause a 2–8 percentage point decline in model performance on certain reasoning-intensive tasks.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/564425API
curl https://kongchang.com/api/v1/knowledge/claims/564425MCP
get_claim(id=564425)