gr-competitor

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

• Third-party content exposure detected (high risk: 0.92). Required workflow uses actionbook to fetch and extract readable text from user-supplied competitor URLs (public web pages like landing/pricing/blog/Wayback), which then gets synthesized by the LLM—an outsider-authored free-text injection path.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires installing and running the actionbook CLI (npm / npx) and explicitly links to https://github.com/actionbook/actionbook and https://actionbook.dev, which are remote code sources fetched/executed during setup/runtime (npm install / npx), so they are runtime external dependencies that execute remote code.