firecrawl-website-design-clone

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md:19-46 describes runtime Firecrawl scraping of a user-supplied URL (public web content) and collecting page markdown/metadata/links and images/screenshot evidence, which would be ingested into the agent’s LLM context for synthesis into DESIGN.md (outsider-authored free text from the scraped page).

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill runs firecrawl at runtime to fetch and ingest a target website and explicitly uses the scraped content to drive the agent's output (example site cited: tesla.com/cybertruck), so remote page content directly controls prompts and is a required dependency.