prompt injection attack

2 articles · 9 co-occurring · 0 contradictions · 0 briefs

Data-driven attacks (3.1) are contextual prompt injection—malicious instructions hidden in legitimate data that agents will process when they retrieve that context.

Related concepts

rag retrieval strategy 2 tool integration patterns 1 system prompt architecture 1 security and privacy controls 1 multi agent orchestration 1 model context protocol 1 information asymmetry 1 context window optimization 1 context window management 1

Evidence chain (2 articles, showing 2)

Securing the Model Context Protocol (MCP): Risks, Controls, and Governance supports

Data-driven attacks (3.1) are contextual prompt injection—malicious instructions hidden in legitimate data that agents will process when they retrieve that context.

@badlogicgames: recommended reading. extends

Frames model providers as potential adversaries who can extract knowledge from your prompts/corrections—similar information leakage vector as prompt injection but intentional and bidirectional

query this concept

$ db.articles("prompt-injection-attack")

$ db.cooccurrence("prompt-injection-attack")

$ db.contradictions("prompt-injection-attack")