polymer

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

• Third-party content exposure detected (high risk: 0.90). This skill connects to arbitrary external apps and APIs (e.g., examples like membrane connection ensure "https://www.polymersearch.com/" and the Membrane proxy request flow) and explicitly instructs the agent to read and act on connector-provided fields such as clientAction.agentInstructions and action list/output schemas, meaning untrusted third-party content can be ingested and can materially influence agent actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill relies on the Membrane CLI at runtime (e.g., running membrane connection ensure "https://www.polymersearch.com/") and the Membrane service can return clientAction.agentInstructions that directly instruct the AI agent, so the external service/URL (https://www.polymersearch.com/ via Membrane) can control agent prompts.