Verified65% confidenceFactExact time
The model itself never runs code; it only writes out requests to call tools, and it is external code that actually executes the tools
3
Sources
65%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAnthropic'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.79% 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.78% similarUnverifiedThe language model itself cannot execute code and can only output text; what actually performs actions is the external harness program.74% 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% similarUnverifiedThe Claude model itself does not execute any tools; the actual execution of tool functions is performed by the developer's Python code, and the permission boundaries and parameter validation are controlled by the developer.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/977670API
curl https://kongchang.com/api/v1/knowledge/claims/977670MCP
get_claim(id=977670)