image-reference-guide

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

• Third-party content exposure detected (high risk: 0.80). This skill's Step 1 ("웹 검색으로 실제 도표 탐색") and the image collection/process sections explicitly require searching, fetching and recording images and URLs from public third‑party sites (e.g., MarketsandMarkets, Statista, government and arbitrary URLs) which the agent is expected to read and use to generate prompts and outputs, thereby exposing it to untrusted external content that could influence actions.