web-research-assistant

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

• Third-party content exposure detected (low risk: 0.10). The workflow performs web searches via the BrowserAct API and ingests arbitrary internet search results and snippets into the agent context.

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

• Potentially malicious external URL detected (high risk: 0.70). The script makes operational HTTPS requests to an arbitrary third-party endpoint (https://mcp.browseract.com/) at runtime to retrieve search results and behavior-influencing content, representing an unknown or third-party operational dependency with weak provenance.