Unverified50% confidenceFactExact time
Pure latent space reasoning cannot call external tools in the middle of reasoning, because the model does not generate tool-call instructions in text form.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedNot all models or platforms support custom tool calls, and in unsupported environments the DFS logic can degrade to a plain-text numbered list interaction.75% similarUnverifiedPrompt injection exploits not a vulnerability in the code parser, but the model's own insufficient awareness of the boundary between instructions and data74% similarUnverifiedA tool's differentiation logic (such as a subagent having dedicated resolution code) is often not written into the official documentation.68% similarUnverifiedEncoder-based decision models do not perform text generation, and their outputs come from a predefined label space, giving them a structural advantage of being hallucination-free68% similarUnverified提示词注入分为直接注入和间接注入两种形式，因语言模型无法区分数据与指令，凡进入上下文窗口的文本都可能被执行68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028113API
curl https://kongchang.com/api/v1/knowledge/claims/1028113MCP
get_claim(id=1028113)