audit-security

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

• Third-party content exposure detected (high risk: 0.85). This skill’s required workflow uses the firecrawl MCP tools to fetch and scrape web pages at runtime (public web content like OWASP Top 10 plus search/CVE results) and ingests their scraped text into the agent context, which can contain outsider-authored free text for indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill issues a runtime scrape of external content via CallMcpTool to "https://owasp.org/Top10/", which will fetch and inject remote page content into the agent's context and thus directly influences its prompts/outputs.