Unverified50% confidenceFactExact time
ReAct架构的核心循环为Thought→Action→Observation，攻击者可以在Observation阶段注入恶意内容
1
Sources
50%
Confidence
Long-term
Relevance
8/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe goal of memory poisoning attacks is to embed malicious instructions in an Agent's persistent memory so that it executes unintended operations in future interactions, which differs from the immediate hijacking of prompt injection.71% similarUnverified推测执行过程中即使指令最终被回滚，其对缓存状态的影响仍会残留，攻击者可通过测量访存时延推断敏感数据70% similarUnverifiedJailbreak techniques can be categorized into prompt injection, adversarial prefix attacks, multi-step context construction, and automated jailbreaking.69% similarUnverifiedReAct范式通过将思考（Thought）→行动（Action）→观察（Observation）三元组交织在同一上下文中构建闭环反馈控制系统69% similarUnverified越狱攻击经历了从简单角色扮演（如DAN提示）到Base64编码绕过、多轮对话渐进式诱导（Crescendo Attack）、对抗性后缀攻击等多个演进阶段68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/787214API
curl https://kongchang.com/api/v1/knowledge/claims/787214MCP
get_claim(id=787214)