#how do i disable verbose mode?
1 messages · Page 1 of 1 (latest)
That’s almost always the session-level /verbose (and sometimes /reasoning) toggle getting turned on, which makes OpenClaw forward tool-call bubbles (and in full, the tool outputs too) into the chat.
In the same Discord thread/channel where it’s spamming you, send:
/verbose off
/reasoning off
If it keeps coming back:
- Open the Control UI → Sessions for that chat and set Verbose = inherit (not
on/full).
(/verbose offstores an explicit override; “inherit” clears it.) - Check your config defaults: make sure you’re not setting
verboseDefault: "on"anywhere. The default lives under:agents.defaults.model.verboseDefault(should be"off")
Docs: https://docs.openclaw.ai/tools/thinking (covers /verbose / /reasoning and what “full” does)
If you paste openclaw --version + whether /verbose says on/full when you type just /verbose, we can narrow down why it flipped on.
thanks that did fix it. i turned verbose off and now the status line shows:
Runtime: direct · Think: low · elevated
what does the elevated marker mean?
That “elevated” marker means the session is running with Elevated Mode enabled.
Practically, Elevated Mode is about where exec runs and how strict approvals are:
- When the agent is sandboxed,
/elevated on|askforces exec to run on the gateway host (instead of inside the sandbox), while still keeping approval prompts when needed. /elevated fullis the “dangerous” one: runs on the host and auto-approves exec (skips approvals).- If your agent isn’t sandboxed, “elevated” may be mostly a status/gating flag (less effect), but it still indicates the session is allowed to do host-level exec.
To see what it’s set to, send:
/elevated
To turn it off for this session: