seo-geo

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to fetch and parse arbitrary public webpages and indexes (SKILL.md Step 1 and curl examples) and includes runtime scripts (e.g., scripts/seo_audit.py, DataForSEO-based scripts, and references to WebSearch/twitter/reddit skills) that ingest open web and social content which the agent is expected to read and use to drive SEO/GEO decisions, creating a clear vector for indirect prompt injection.