multi-llm-mcp-server

SUSPICIOUS. The skill’s stated purpose matches its high-level capabilities, and the install sources appear to be official ecosystems rather than overtly malicious payloads. However, it requests many provider credentials, can route data to multiple external LLMs, and can trigger Codex with full system access. Because the actual LLM_MIX.py code is not provided, data-flow integrity and credential forwarding cannot be verified, so the risk is medium rather than benign.