Concept graph
Short definitions with deeper context and cross-links to sibling terms.
Agent security
Indirect prompt injection happens when untrusted external content, such as a webpage, email, document, or tool result, contains instructions that try to steer an AI system.
