grad-agenda-setting

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md methodology (Step 1: "Define the media agenda (content analysis of coverage frequency/prominence)") and the example scenario explicitly require ingesting and analyzing open/public third‑party sources (e.g., media outlets like 聯合報/自由時報, Meltwater feeds, and social media/survey data), so the agent is expected to read untrusted web/social content that could contain instructions influencing analysis and actions.