ai-persistence

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow for state persistence (withPersistence / withGenerationPersistence) reads and later reconstructs conversation state/messages and artifact metadata from caller-controlled request identifiers (threadId, runId, artifactId) via store methods like loadThread/reconstructChat/retrieveArtifact, which can expose the agent to outsider-authored free text if an outsider can submit that state to the backend queue/feed being monitored/consumed.