product-appeal-analyzer

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs fetching and analyzing live landing-page URLs (see SKILL.md "Self-Contained Tools" step 2: "WebFetch the target URL to analyze live content") and the scripts/appeal_scorer.py accepts a , meaning the agent ingests arbitrary public web pages (untrusted third‑party content) that could carry indirect instructions influencing its analysis and actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly says it will "WebFetch the target URL to analyze live content" and the runner/script accepts a user-supplied URL (e.g., the target landing-page URL provided at runtime), so fetching arbitrary external pages will be injected into the agent's analysis flow and can directly control prompts—flagging the user-supplied target URL (any external landing-page URL) as a runtime dependency.