pmax-launch-kit

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

• Third-party content exposure detected (high risk: 0.95). At runtime the skill performs WebFetch/WebSearch on user-supplied URLs and public web pages, ingesting scraped page text and search result snippets into the agent’s LLM context (outsider-authored free text from public web content).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly WebFetches the user-supplied brand URL at runtime (e.g., "cogny.com" or any provided URL) and uses the fetched site content and WebSearch results to directly drive the agent's prompts and outputs, so the external page content can control the agent's behavior.