pr

The skill’s main workflow is coherent and does not show credential exfiltration or a suspicious install path. The request to read model identity from the system prompt is a security concern; autonomous mode also enables GitHub side effects without per-action approval. Overall suspicious rather than confirmed malicious.