minding

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests outsider-authored free text when it performs fact-level recall and the cross-realm sweep via nks_search(realm="@<handle>/mind", q=<term>)/agenda reads across cards populated by user-facing content, meaning user/outsider submissions into the underlying mind graph can be retrieved and consumed by the LLM.