ai-seo

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to test queries and read AI search results and third‑party pages (e.g., "Test 10-20 of your most important queries across platforms" naming ChatGPT, Perplexity, Google AI Overviews, and later references to Wikipedia/Reddit and checking robots.txt), meaning the agent is expected to fetch and interpret untrusted public web content that can change its actions.