netbox-assurance

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

• Third-party content exposure detected (high risk: 0.75). Assurance deviations and field-level before/after change details are produced from runtime-ingested “actual network state” coming from configured data sources (e.g., Discovery agents/controllers/Diode SDK), and those deviation texts are then presented in the NetBox Assurance UI and thus become LLM-readable context for the agent; this is outsider-authored content because it originates from external network/controller data rather than the operating user’s own authored prompts/files.