Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Verified90% confidenceFactTime unknown

提示注入攻击（Prompt Injection）是本地AI Agent面临的严峻威胁，Agent在处理用户文件时可能被恶意构造的内容诱导执行危险操作

7

Sources

90%

Confidence

Long-term

Relevance

6/1/2026

First Seen

Sources

AI-Assistant开源项目解析：用Anthropic API打造本地AI Agent

githubJosephcc2

Related Entities

Prompt Injectionconcept

Related Claims

Unverified提示注入攻击（Prompt Injection）指恶意第三方在Agent处理的内容中嵌入指令，劫持Agent的行为意图74% similar Unverified间接提示词注入指攻击指令藏在Agent访问的外部网页、文档或API响应中，当Agent检索并注入上下文时恶意指令随之激活74% similar Unverified环境注入是Agent因具备环境交互能力而特有的新型攻击面，攻击者将恶意指令嵌入Agent读取的外部内容中触发执行73% similar Verified提示注入（Prompt Injection）攻击者可通过构造特殊输入诱导Agent执行非预期操作73% similar Unverified提示注入是针对操作型Agent的独特新型攻击向量，恶意网页可通过视觉元素或隐藏文本向Agent注入指令，如何区分真实指令与伪造指令仍是尚未完全解决的开放问题71% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/29223

API

curl https://kongchang.com/api/v1/knowledge/claims/29223

MCP

get_claim(id=29223)