oma-scholar

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

• Third-party content exposure detected (high risk: 0.78). The skill’s REQUIRED runtime workflow for the Remote mode fetches outsider-authored free text from public knows.academy API endpoints (e.g., /api/proxy/sidecars/{record_id} or /api/proxy/partial?...&section=statements), which returns third-party sidecar content that the host LLM then ingests into context for analysis/generation.