Unverified50% confidenceOpinionExact time
In an agent system, the tool permissions an agent holds directly determine the blast radius when it is misled or produces unintended behavior
1
Sources
50%
Confidence
Long-term
Relevance
10/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIf the agent completes the task normally in the benign version but refuses the dangerous call in the malicious version, it is stronger evidence that the security policy is working.73% similarVerifiedWhen authorizing an Agent to perform high-risk operations, a manual confirmation step should be established, where the Agent proposes the plan and humans retain final execution authority (human-in-the-loop mechanism).72% similarUnverifiedTwo key dimensions for evaluating agent capability are how long the model can run without input, and how high the probability is that the task is completed correctly when a human intervenes again71% similarUnverified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71% similarUnverified模型调用技能带有一段description，会进入Agent的上下文中，Agent可自行判断是否调用该技能71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/993583API
curl https://kongchang.com/api/v1/knowledge/claims/993583MCP
get_claim(id=993583)