fact-check-loop

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests document text extracted from a user-provided file path and also retrieves outsider-authored source URLs (via a web-fetch tool) that the user supplies, and both are read by the fact-check subagent at runtime.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs installing and running the external converter markitdown at runtime (https://github.com/microsoft/markitdown), which fetches remote code and is executed via pip install and python3 -m markitdown for document conversion.