deploy-pythonanywhere

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md the agent at runtime uploads user-provided application code and then performs reads of provider-side logs/error output (e.g., /var/log/.pythonanywhere.com.error.log), which can contain outsider-authored free text generated during the app’s handling of user requests.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's user setup step explicitly runs "curl -fsSL | bash" (SKILL.md line 24), which fetches remote content and pipes it to a shell, thus executing remote code at runtime via the provided URL "".