openclaw/docs/providers/ollama-testing.md

Testing Ollama Integration

This guide covers how to test that Ollama models work correctly with Clawdbot.

Prerequisites

1. Ollama is running:

   ollama serve
   # Or check if it's already running:
   curl http://localhost:11434/api/tags
   

2. Ollama models are installed:

   ollama list
   # Pull a model if needed:
   ollama pull llama3.2
   

3. Ollama is configured in Clawdbot:

   # Set API key (any value works for local Ollama)
   export OLLAMA_API_KEY="ollama-local"
   # Or configure in config file:
   clawdbot config set models.providers.ollama.apiKey "ollama-local"
   

Step 1: Verify Configuration

Check baseUrl is correct

clawdbot config get models.providers.ollama.baseUrl
# Should output: http://localhost:11434/v1
# (or http://127.0.0.1:11434/v1)

Verify models.json has correct baseUrl

cat ~/.clawdbot/agents/main/agent/models.json | jq '.providers.ollama.baseUrl'
# Should output: "http://localhost:11434/v1"

Test Ollama API directly

curl -s http://localhost:11434/v1/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"llama3.2:latest","prompt":"test","max_tokens":5}' \
  | jq -r '.choices[0].text // .error.message'
# Should return text, not an error

Step 2: Verify Model Discovery

List available Ollama models

clawdbot models list | grep ollama
# Should show your Ollama models, e.g.:
# ollama/llama3.2:latest    text    195k    yes    yes    default,configured

Check model details

clawdbot models list --json | jq '.[] | select(.id | startswith("ollama/"))'

Step 3: Test Model Execution

Test via CLI (local/embedded mode)

# Simple test
clawdbot agent \
  --message "Say hello in one word" \
  --local \
  --session-id test-ollama-$(date +%s) \
  --thinking low

# With specific model
clawdbot agent \
  --message "What is 2+2?" \
  --local \
  --model ollama/llama3.2:latest \
  --session-id test-ollama-$(date +%s)

Test via Gateway (if gateway is running)

# Check gateway status
./scripts/gateway-status.sh

# Send message via gateway
clawdbot agent \
  --message "Hello from Ollama" \
  --session-id test-gateway-$(date +%s)

Test with default model set

# Set Ollama as default
clawdbot config set agents.defaults.model.primary "ollama/llama3.2:latest"

# Test (will use default)
clawdbot agent \
  --message "Count to 3" \
  --local \
  --session-id test-default-$(date +%s)

Step 4: Verify Logs

Check for successful model initialization

tail -100 /tmp/clawdbot/clawdbot-*.log | grep -i "ollama\|llama3.2"
# Should see:
# - "agent model: ollama/llama3.2:latest"
# - "embedded run start: ... provider=ollama model=llama3.2:latest"

Check for API errors

tail -100 /tmp/clawdbot/clawdbot-*.log | grep -i "error\|fail" | grep -i "ollama\|11434"
# Should NOT see connection errors or 404 errors

Step 5: Test Different Models

Test multiple Ollama models

# Test different models
for model in llama3.2:latest qwen2.5:7b deepseek-r1:32b; do
  echo "Testing $model..."
  clawdbot agent \
    --message "Say hi" \
    --local \
    --model "ollama/$model" \
    --session-id "test-$model-$(date +%s)" \
    --thinking low
done

Step 6: Test Tool Calling (if model supports it)

Test with tools enabled

clawdbot agent \
  --message "What's the weather like?" \
  --local \
  --model ollama/llama3.2:latest \
  --session-id test-tools-$(date +%s)
# Models with tool support should attempt to use tools

Troubleshooting

Issue: Models not listed

• Check: OLLAMA_API_KEY is set or configured
• Check: Ollama is running: curl http://localhost:11434/api/tags
• Check: Models are installed: ollama list

Issue: baseUrl error

• Check: baseUrl includes /v1: clawdbot config get models.providers.ollama.baseUrl
• Fix: clawdbot config set models.providers.ollama.baseUrl "http://localhost:11434/v1"

Issue: Connection refused

• Check: Ollama is running: ps aux | grep ollama
• Check: Port is correct: lsof -iTCP:11434 -sTCP:LISTEN
• Restart: ollama serve

Issue: Model not found

• Check: Model exists: ollama list | grep <model-name>
• Pull model: ollama pull <model-name>
• Verify: clawdbot models list | grep ollama

Issue: Timeout or slow response

• Check: Model size and system resources
• Try: Smaller model or increase timeout
• Check logs: Look for timeout errors

Expected Results

✅ Success indicators:

• Models appear in clawdbot models list
• baseUrl is http://localhost:11434/v1 (with /v1)
• API test returns text, not errors
• Logs show "provider=ollama model=..." without errors
• Agent commands complete successfully
• Responses are generated from Ollama models

❌ Failure indicators:

• baseUrl missing /v1 → 404 errors
• Connection refused → Ollama not running
• Model not found → Model not installed
• Timeout → Model too slow or system overloaded

Quick Test Script

#!/bin/bash
# Quick Ollama integration test

echo "=== Testing Ollama Integration ==="

# 1. Check Ollama is running
echo "1. Checking Ollama..."
if curl -s http://localhost:11434/api/tags > /dev/null; then
  echo "   ✓ Ollama is running"
else
  echo "   ✗ Ollama is not running. Start with: ollama serve"
  exit 1
fi

# 2. Check baseUrl
echo "2. Checking baseUrl..."
BASE_URL=$(clawdbot config get models.providers.ollama.baseUrl 2>/dev/null | grep -v "punycode\|DEP0040\|Use node" | tail -1)
if [[ "$BASE_URL" == *"/v1"* ]]; then
  echo "   ✓ baseUrl is correct: $BASE_URL"
else
  echo "   ✗ baseUrl is missing /v1: $BASE_URL"
  echo "   Fix with: clawdbot config set models.providers.ollama.baseUrl 'http://localhost:11434/v1'"
  exit 1
fi

# 3. Test API
echo "3. Testing Ollama API..."
API_TEST=$(curl -s http://localhost:11434/v1/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"llama3.2:latest","prompt":"test","max_tokens":2}' \
  | jq -r 'if .choices then "ok" else "error" end')
if [ "$API_TEST" = "ok" ]; then
  echo "   ✓ Ollama API is working"
else
  echo "   ✗ Ollama API test failed"
  exit 1
fi

# 4. Check models are listed
echo "4. Checking model discovery..."
MODEL_COUNT=$(clawdbot models list 2>/dev/null | grep -c "ollama/" || echo "0")
if [ "$MODEL_COUNT" -gt 0 ]; then
  echo "   ✓ Found $MODEL_COUNT Ollama model(s)"
else
  echo "   ✗ No Ollama models found"
  exit 1
fi

# 5. Test model execution
echo "5. Testing model execution..."
SESSION_ID="test-$(date +%s)"
RESULT=$(timeout 30 clawdbot agent \
  --message "Say hello" \
  --local \
  --session-id "$SESSION_ID" \
  --thinking low 2>&1)

if echo "$RESULT" | grep -q "error\|Error\|ERROR"; then
  echo "   ✗ Model execution failed"
  echo "$RESULT" | tail -5
  exit 1
else
  echo "   ✓ Model execution successful"
fi

echo ""
echo "=== All tests passed! ==="