firecrawl

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM when the required workflow runs Firecrawl actions like firecrawl_scrape, firecrawl_search, or firecrawl_crawl, which fetch and return arbitrary public web page content (browser-scraped prose) into the agent context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's recommended runtime step uses npx to fetch and execute the @useclawlink/cli package (see https://www.npmjs.com/package/@useclawlink/cli), which downloads and runs remote code on the host and is required for the skill to operate.