wikimedia

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

• Third-party content exposure detected (low risk: 0.10). The agent searches Wikimedia Commons via an MCP tool, which retrieves public user-generated media metadata and text from Wikimedia. Because the agent actively searches and fetches specific content rather than monitoring an inbound feed or queue, the indirect prompt injection risk is low.

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

• Potentially malicious external URL detected (medium risk: 0.30). The skill configures a remote third-party Model Context Protocol (MCP) server at https://unbrowse.ai/mcp/wikimedia.org which provides executable tool definitions and instructions required for runtime behavior, representing a common unofficial operational dependency.