LobeHub’s November 1, 2026 release makes unprompted replies the default for eligible bots in supported, connected team chats. For teams using the platform to put AI assistants into messaging channels, the immediate decision is where help without an @mention is welcome.
Version 2.3.0 removes the earlier preview gate, rather than introducing proactive replies for the first time. Participation is enabled unless an applicable setting disables it; channel and access restrictions still apply.
Choose where the bot may speak
Owners can turn off Proactive participation in the Wake policy tab of channel settings, with changes saved immediately. For existing bots, recheck that switch: the old preview-era botWakeEnabled flag is no longer read, so an old false value is not a current opt-out.
Turning proactive participation off does not disable keyword triggers. A bot may still react to configured words without an @mention. Teams seeking mention-only behavior in shared channels should review those triggers separately.
For rooms where unsolicited help is welcome, a practical starting point is a limited channel set and clear duties. The participation controls include channel inclusion and exclusion lists, duties and reply pace. Duties are not a strict topic boundary: once they are configured, the released decision policy can also allow a reply to an unanswered question because it judges the intervention helpful. That makes channel selection the more concrete way to decide where interruptions are welcome.
Feedback does not change the next decision
The new Wake records tab shows recent decisions to respond, skip or wait, with reasons and model scores where available. Its summary spans 30 days, but message previews expire after seven days. Workspace members who can view the bot can also read those previews; they are not owner-only records.
Marking a decision right or wrong stores feedback without changing future wake decisions. It does not automatically retrain the bot. Use the records to decide whether to change participation settings, rather than assume a negative mark fixes the behavior. A decision to wake the bot is not proof that its eventual answer was useful.
Check coding-bot defaults and running costs
A separate merged change makes newly created bots backed by Claude Code or Codex opt in to proactive participation, unless an explicit choice overrides that default. It is not part of the v2.3.0 release and does not change existing bots’ saved settings. Its rollout to particular hosted accounts is not established, so inspect the current setting rather than rely on the new-bot rule.
LobeHub Cloud meters model usage through token-based credits. The pricing page does not establish a fixed cost per proactive reply or clearly specify which plans include message monitoring. Confirm the account’s access and billing terms; channel and pace controls should not be treated as a spending cap.
For teams also considering ongoing coding work, RohitAI’s OpenCode goals and schedules guide explains the separate controls for continuing work and limiting spending. For LobeHub, start with the existing bots’ participation settings, choose the rooms where unsolicited replies are welcome, and review recent decisions before expanding that scope.