Natural Moderation

When a Watcher warns someone, it's phrased in natural language consistent with its persona, rather than a fixed, robotic template — while staying consistent with your community's policy and the graduated-enforcement rule described in "Moderation overview."

Example: "Hey @username, that's the second promotional post I've had to remove. Please keep promotional links out of this community — another violation may result in further action."

A warning never restates the specific violating content back into the chat, and it never sounds like a generic canned message repeated word-for-word every time — the phrasing varies while the underlying enforcement decision stays consistent and rule-based.