Unverified50% confidenceFactExact time
When prompt injection is hidden in tool output it is called "indirect prompt injection" and is more dangerous, because developers often guard only against user input and trust tool returns.
1
Sources
50%
Confidence
Long-term
Relevance
10/21/2026
First Seen
Sources
Don't Rush to Deploy AI Agents: A Complete Design Guide to Tool Use and Function Calling
youtubeSystems Architecture Hub10/21/2026
Related Entities
Related Claims
UnverifiedIndirect prompt injection infiltrates through channels such as tool return content and retrieved documents, posing a greater threat to agentic systems because agents trust tool output by default.80% similarUnverifiedPrompt injection refers to an attacker embedding instructions in content that is treated as data, inducing the model to execute beyond its authority, and its root cause is the system's failure to isolate untrusted input from the trusted instruction channel.75% similarUnverifiedHidden settings leaks often occur because developers directly expose options in the client that should be controlled server-side, and users can discover them through browser developer tools or network packet capture tools74% similarVerifiedIndirect prompt injection attacks plant hidden instructions within normal business data to induce Agents into executing unauthorized operations, representing a security risk at the semantic level rather than the code level.73% similarUnverifiedPrompt injection or abnormal user input may allow an Agent to access data that should have been blocked or perform operations that should not have been executed, which is an architectural-layer vulnerability where permission checks are bypassed.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1080449API
curl https://kongchang.com/api/v1/knowledge/claims/1080449MCP
get_claim(id=1080449)