crawling-social-media

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow takes a user-provided social media URL, creates a crawl task, and then returns extracted title/description/tags—those fields are derived from outsider-authored social media post content fetched at runtime (public web content), which is then ingested into the agent’s LLM context via the crawl results.