feedback

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow reads outsider-authored free text only from the user calling /feedback and then ingests it into the agent’s own feedback realm via its search/placement steps (nks_semantic_search(realm="@nks/feedback", q=<the case as one sentence>)), which is a direct injection path from an untrusted external author.