social-creative-builder

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime reads include “destination URL (User-provided)” and it explicitly allows “Public open-web checks” via connectors (e.g., scripts/connectors/tavily.py / bluesky.py) and treats “scraped page” text as untrusted input; if that fetched page/body text is ingested into the LLM context, it is outsider-authored free text from public web content.