research-account

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s Step 2 ingests outsider-authored free text at runtime by letting the user supply target URLs/domains for parallel.extract (“read from their own pages”) and job results from theirStack.searchJobs (“searchJobs”).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill issues a runtime call to the "parallel.extract" connector that fetches web pages listed in the records (e.g., https://acme.com and https://acme.com/careers on runtime) which will inject remote webpage content into the agent's context and thus directly influence its generated briefing.