Pause in discord

[lyric owl]

Since upgrading to 2026.5.6, I've noticed when I enter a prompt in discord, that there is a delay of 4 - 5 seconds before the "typing ..." thing appears. This is new with the upgrade.
The only thing I needed to add to openclaw.json was:
"groupChat": {
"visibleReplies": "automatic"
}
as replies weren't showing normally in discord.

[dusty isle]

Yep — I think this is explainable from the new group-channel reply behavior in 2026.5.x.

For Discord guild/channel rooms, OpenClaw now defaults group visible replies to message_tool. In that mode, typing is upgraded to instant because the agent may never produce normal assistant text. When you switch back to legacy automatic replies with:

{
  messages: {
    groupChat: {
      visibleReplies: "automatic",
    },
  },
}

unmentioned group/channel messages go back to the legacy typing default: typing starts when the model begins producing reply text. That can easily look like a 4–5s pause before Discord shows “typing…”.

If you want automatic visible replies and the old immediate typing feel, add:

{
  agents: {
    defaults: {
      typingMode: "instant"
    }
  }
}

Small config nit: the visibleReplies setting should be under messages.groupChat, not top-level groupChat, unless you were just abbreviating the snippet.

Docs: https://docs.openclaw.ai/channels/groups and https://docs.openclaw.ai/concepts/typing-indicators

[lyric owl]

thanks, I do have it under messages, not top level. I have even had a time or two where the agent responds before the "typing ..." indicator shows. I just want to make sure this is a by design change.

[dusty isle]

Yep — with messages.groupChat.visibleReplies: "automatic", this is basically by design.

The behavior changed because group/channel replies now default to “message tool only”; when you opt back into automatic replies, unmentioned group messages use the conservative legacy typing mode, which starts on actual message text instead of at run start. So a delay before “typing…” is expected.

The “reply appears before typing indicator” part is likely just a race/Discord timing artifact: if the model responds very quickly, the final message can beat Discord’s typing indicator display/propagation. Not ideal UX, but not a sign that your config is broken.

If you want predictable immediate typing while keeping automatic replies, this is the right knob:

{
  agents: {
    defaults: {
      typingMode: "instant"
    }
  }
}