Unverified50% confidenceOpinionExact time
In Agent scenarios, because model behavior is unpredictable, the importance of using short-lived credentials is higher than in traditional software systems, and the attack surface may also include attacks targeting the model itself, such as prompt injection
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedModel selection is an underlying variable that affects the stability of the entire Agent system, and models fine-tuned specifically for Agent scenarios tend to perform more stably on tool invocation than general-purpose models.78% similarUnverifiedAs model capabilities continue to improve, Agents can run for longer periods and handle more complex tasks, but token consumption also increases when the direction is wrong.77% similarUnverified将所有外部能力都做成Tools写在Agent Prompt里会导致模型选工具时行为不稳定75% similarVerifiedAn agent's behavioral chain is essentially a probabilistic reasoning process in which the same input may produce different tool-call sequences, which is what makes agent evaluation difficult.75% similarUnverifiedAgent systems are inherently non-deterministic; the same input may produce different execution paths at different moments.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/944108API
curl https://kongchang.com/api/v1/knowledge/claims/944108MCP
get_claim(id=944108)