Unverified50% confidenceFactExact time
Anthropic's documentation explicitly states that the model itself never performs any operation, it only writes down requests, and what actually executes is the host application code.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe model itself never runs code; it only writes out requests to call tools, and it is external code that actually executes the tools79% similarUnverifiedSystem Prompt与用户输入本质上都是模型的输入token，模型并不存在独立的规则执行引擎来强制保证指令被遵守68% similarUnverifiedOpenAI reported that after compression the model continued to complete the task with no mention of the appended instruction, and no behavioral difference attributable to the self-generated instruction was observed.68% similarUnverifiedIn tool calling, the model itself does not execute code but is only responsible for deciding which tool to call, while actual execution authority remains in the host program controlled by the developer.64% similarUnverifiedIn Ollama, you can use read-only commands to view model configurations, including the template, system prompt, and stop settings, without modifying the model.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1009205API
curl https://kongchang.com/api/v1/knowledge/claims/1009205MCP
get_claim(id=1009205)