openclaude-multi-llm

SUSPICIOUS. The skill’s core purpose and capabilities mostly align: it is explicitly a shim that reroutes Claude Code to OpenAI-compatible providers. The main risks are proportional but real: it installs an external CLI, forwards API keys and possibly Codex auth material to that CLI, permits arbitrary provider endpoints, and uses a less-verifiable custom git host for source installs instead of the project’s clearer GitHub origin. No confirmed malware, covert behavior, or pre-execution abuse is present, but the trust and credential-routing model is broader than a normal single-vendor integration.