Unverified50% confidenceFactExact time
The model never directly touches databases, APIs, or files; it issues structured tool requests, which the orchestrator validates, executes, and returns results for, and the model can only reach the outside world through the capabilities the application chooses to expose.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEngineering practices based on the assumption that a model has been compromised include: independently validating model outputs, applying the principle of least privilege to restrict the data and tools accessible to the model, continuously monitoring and auditing inputs and outputs, and isolating scenarios that involve ingesting external content.71% similarUnverifiedFunction calling adopts a separation design where "the model is responsible for understanding intent and the app is responsible for executing operations," ensuring the model cannot directly access the database.69% similarVerifiedThe basic principle of tool calling is that the model identifies scenarios requiring external capabilities, outputs a structured function call request (usually in JSON format), and after the external system executes it, the result is returned to the model to continue reasoning.69% similarUnverified普通大模型的工具调用是被动的，模型只能声明想调用某个函数及其参数，实际执行仍需外部系统完成68% similarUnverifiedThe model parameter in an API request is used to specify which model to use.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1007070API
curl https://kongchang.com/api/v1/knowledge/claims/1007070MCP
get_claim(id=1007070)