Based on your questions, here's what's happening and how to address it:

## Issue 1: "qwen3-coder-agent" does not support thinking

The error occurs because OpenHands is sending thinking-mode parameters to the

qwen3-coder-agent

model, but that model variant doesn't support thinking mode. Here's how to fix it:

### Option A: Use a model that supports thinking

The

qwen3-coder:30b

base model doesn't have thinking enabled by default. You need to either:

- Use a thinking-enabled variant like

qwen3-coder:30b-abliterator

or similar

- Or disable thinking mode in OpenHands

### Option B: Disable thinking mode in OpenHands

Add this to your Modelfile:

```dockerfile

PARAMETER thinking off

```

Or in your

config.toml

or environment settings, look for a

thinking

or

enable_thinking

setting and set it to

false

.

---

## Issue 2: Viewing System Prompt in OpenHands

You can view the system prompt being sent to the LLM by:

### Method 1: Check the OpenHands Docker Container

```bash

docker exec -it <container_name> bash

# Then look at the config or logs

cat /app/openhands/config.py | grep -i system

```

### Method 2: Enable debug logging

Set environment variable in your docker-compose:

```yaml

environment:

LOG_LEVEL: DEBUG

```

Then check logs:

```bash

docker logs <container_name> 2>&1 | grep -i "system.prompt\|system_prompt\|prompt"

```

### Method 3: Find the system prompt in source code

The system prompts for OpenHands are typically defined in:

-

/app/openhands/llm/

directory

- Look for files like

prompt.py

or

system.py

```bash

docker exec -it <container_name> find /app -name "*.py" -exec grep -l "system.*prompt\|SYSTEM_PROMPT" {} \;

```

---

## Quick Fix for Your Error

The most likely fix is to either:

1. Add to your Modelfile:

PARAMETER disable_thinking true

or

PARAMETER thinking false

2. Use a different model identifier that doesn't trigger thinking mode

3. Check OpenHands settings in the UI for a "Thinking Mode" toggle (sometimes it's under advanced settings)

Would you like me to help you locate the specific configuration file in your OpenHands Docker setup to disable thinking mode?

docker run -it --rm --pull=always \
  -e AGENT_SERVER_IMAGE_REPOSITORY=<http://ghcr.io/openhands/agent-server|ghcr.io/openhands/agent-server> \
  -e AGENT_SERVER_IMAGE_TAG=1.19.1-python \
  -e OH_AGENT_SERVER_ENV='{"LLM_REASONING_EFFORT": "none"}' \
  -e LOG_ALL_EVENTS=true \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v ~/.openhands:/.openhands \
  -p 3000:3000 \
  --add-host host.docker.internal:host-gateway \
  --name openhands-app \
    docker.openhands.dev/openhands/openhands:1.7

reasoning_effort=high   -> think: True  (rejected)
reasoning_effort=none   -> think: False (ALSO rejected)
reasoning_effort=null   -> no think param (works!)
drop_params=True + high -> think: True  (no help - litellm thinks param is supported)

LLM_DROP_PARAMS

think