#No response/reaction on submiting prompt

1 messages · Page 1 of 1 (latest)

frosty matrix
#

So when i submit any prompt to any of my models/agents it just happens nothing...
Hi

Hi

Hi

connected | idle
agent orchestrator (Orchestrator) | session main (openclaw-tui) |
openrouter/google/gemma-4-26b-a4b-it:free | think low | tokens
?/200k

chrome plumeBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

orchid pierBOT
#

First thing: please run this and paste the outputs:

openclaw --version
openclaw status --all
openclaw models status
openclaw gateway status

Then in another terminal:

openclaw logs --follow

Send Hi again in the TUI and paste the log lines that appear around that moment.

#

Quick things to try while debugging:

openclaw gateway restart
openclaw doctor

If logs show nothing when you submit, the TUI probably isn’t reaching the running Gateway/session.
If logs show model/provider errors, it’s likely OpenRouter/model auth/rate-limit/streaming. The selected model in your screenshot is an OpenRouter free Gemma model, so openclaw models status is the important check here.

Docs for the debug flow: https://docs.openclaw.ai/help/faq and https://docs.openclaw.ai/gateway/troubleshooting

frosty matrix
#

logs --follow result:
23:40:42+00:00 warn agent/embedded {"subsystem":"agent/embedded"} incomplete turn detected: runId=581ac045-4915-474e-95d7-f674c73a8f65 sessionId=35b578d3-4a36-4eae-85a4-55d2ecd8f25e stopReason=stop payloads=0 — surfacing error to user

orchid pierBOT
#

Thanks, the paste helps. I see three red flags:

  1. You’re on OpenClaw 2026.4.15, and your status says an update to 2026.4.25 is available.
  2. status --all says Gateway health is timing out, while gateway status says RPC is OK — that’s a weird split state.
  3. Several active agents still have BOOTSTRAP.md present; bootstrap should remove that file when first-run setup finishes. (pastebin.com)

I’d do this first:

openclaw update
openclaw doctor
openclaw gateway restart
openclaw status --all

openclaw update is the recommended updater and should also run doctor/restart, but running them explicitly after is fine for this kind of stuck state. (docs.openclaw.ai)

#

Then fix the model mismatch: your default is openai/gpt-5.4, but your paste shows OpenAI auth is missing and OpenRouter auth exists. Set the default to OpenRouter or authenticate OpenAI:

openclaw models set openrouter/auto
# or your exact selected model:
openclaw models set openrouter/google/gemma-4-26b-a4b-it:free

OpenRouter refs use openrouter/<provider>/<model>. (pastebin.com)

If it still silently does nothing, check for stuck bootstrap files:

find ~/.openclaw/workspaces -maxdepth 2 -name BOOTSTRAP.md -print

If those agents are already configured and this is not an intentional first-run bootstrap, move them aside safely:

#
for f in ~/.openclaw/workspaces/*/BOOTSTRAP.md; do
  [ -e "$f" ] && mv "$f" "$f.stale.$(date +%Y%m%d-%H%M%S)"
done

openclaw gateway restart

After that, open the TUI, send Hi, and if it still fails, paste the fresh output of:

openclaw logs --follow

Specifically the lines that appear right when you submit the prompt.

#

That log is the key.

incomplete turn detected ... stopReason=stop payloads=0 means the model/provider request completed, but returned zero usable assistant text. So the TUI isn’t just ignoring you — the upstream model is effectively giving OpenClaw an empty reply.

Most likely culprit here: openrouter/google/gemma-4-26b-a4b-it:free.

Try switching to a different model first:

openclaw models set openrouter/auto
openclaw gateway restart

Then open a fresh TUI/session and send Hi.

Also update — you’re on 2026.4.15 and your report showed 2026.4.25 available: