Function validate_whatsapp_semantics 

pub fn validate_whatsapp_semantics(
    alias: &str,
    wa: &WhatsAppConfig,
) -> Vec<ValidationWarning>

Surface WhatsApp chat-policy keys that are accepted but never consulted.

self_chat_mode is read only by the Web transport inside its mode == Personal block, so under mode = "business" it validates cleanly and has no effect. mode selects ZeroClaw’s policy posture, not a WhatsApp account type: both modes drive the same linked-device session, and the self-chat affordance is scoped to the personal branch by design.

dm_policy and group_policy are consulted under BOTH modes, so they are not reported here.

allowed_groups is separate. It is the group-identity gate and is consulted under both modes, so it is never inert. is_group_chat_allowed returns true for an empty list only under group_policy = "all"; under every other policy an empty list admits no group. That closure is reported below by its own warning rather than as an inert key.

Warnings only, no behaviour change to the validator itself. But be precise about WHICH reliance is reported, because this sentence used to promise more than the function delivers: under mode = "business" the inert-key warning covers self_chat_mode ONLY. dm_policy and group_policy are now live in business mode, so an operator who was relying on the old permissive business behaviour gets NO warning here and can start silently dropping DMs or groups under the default allowlist. That upgrade note belongs in the channel book and the release notes; config validate will not surface it for them.

Called from Config::collect_warnings, so this reaches the CLI and the gateway dashboard on the same path as the other warnings.