content-trend-researcher

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md and HOW_TO_USE explicitly require the agent to analyze and generate outputs based on content from public, user-generated platforms (e.g., Reddit, Medium, YouTube, blogs, Substack, X) — evidence the skill ingests untrusted third‑party content as part of its workflow and that content can influence decisions and next actions.