ti-ingest

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

• Third-party content exposure detected (high risk: 0.75). The skill’s scheduled poll reads raw TI signal files from {OUTPUT_DIR}/inbox/ (i.e., analyst- or externally-supplied signal.json content) and copies the free-text claim/references into the LLM-ingested queue row for downstream processing, which is outsider-authored input at runtime.