enabled, channel metadata (topic/description) can be incorporated into the model's system prompt. Prompt injection is a documented risk for LLM-driven systems. This issue increases the injection surface

OpenClaw: Isolated cron awareness events were recorded as trusted system