Unverified50% confidenceFactExact time
In multi-agent systems, when Agent A calls Agent B, which then calls tool C, the trust assumptions in the intermediate links are often never examined, and each hop can become a channel for attackers to inject instructions.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe dynamic tool selection characteristic of agentic workflows enables agents to call arbitrary APIs or even spawn new agents based on runtime reasoning, causing trust boundaries to expand dramatically.76% 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.74% 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74% similarUnverifiedThe transitive chain of trust is the amplification mechanism of indirect prompt injection: when an Agent uses the output of tool A as input to call tool B, embedded instructions propagate along the chain and may trigger unintended calls.73% similarUnverifiedIn multi-agent architectures, messages sent from one agent to another should be protected in the same way as user input, and each agent should only possess the permissions necessary to run its tools.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1030562API
curl https://kongchang.com/api/v1/knowledge/claims/1030562MCP
get_claim(id=1030562)