Unverified50% confidenceFactExact time
攻击者通过提示词注入控制Agent的决策逻辑后，可利用Agent已获授权的所有工具和接口实现全链路攻击，形成混淆代理人（Confused Deputy）问题
1
Sources
50%
Confidence
Long-term
Relevance
8/12/2026
First Seen
Sources
Related Claims
UnverifiedAn 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% similarVerifiedIn LangChain's Agent architecture, the Agent itself is only responsible for reasoning and generating tool-call instructions, while the Agent Executor is responsible for executing tools, handling responses, and managing the iteration loop.75% similarVerifiedAgent 的核心架构通常由「感知-推理-行动」循环构成75% similarVerifiedThe principle of indirect prompt injection is that an attacker disguises malicious instructions as normal data, delivers them to the model through a data source trusted by the agent, and induces the model to interpret the data as instructions and execute them.74% similarUnverifiedThe generated Agent configuration includes tool permission settings that specify which tools the Agent can use, and supports manual adjustment afterward.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/736773API
curl https://kongchang.com/api/v1/knowledge/claims/736773MCP
get_claim(id=736773)