bridge

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This worker intentionally opens a long-lived outbound WebSocket to a remote iii instance and provides mechanisms to invoke or expose local functions across that channel, which can enable remote code execution or covert data exfiltration if misconfigured or connected to an untrusted endpoint.

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

• Third-party content exposure detected (high risk: 0.70). The required workflow is a configuration-driven bridge that opens a runtime WebSocket to a configured url and exchanges request/response messages; any free-form text contained in those remote messages (outsider-authored by the other iii instance) can be ingested into the agent/LLM context via the bridge call payload/response path.