quarterly-connect

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md and README explicitly require performing web searches of public company pages (e.g., "search for '[Company Name] core values'") to extract and store a company's publicly available core values, and those extracted third-party pages are then used to drive questions, alignments, and final actions—creating a clear path for untrusted public content to influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly performs live web searches during runtime to fetch company core values from external company pages (e.g., company "About/Values" pages such as https://.com/about or other search-result URLs) and injects those findings into prompts and the final report, so external URLs returned by web search directly control agent instructions.