List index out of range msg_i

[sodoherty]

I posted earlier but it’s closed.

Tools are failing with list index out of range when using Ollama in CrewAI. Debugging it appears to be happening in litellm.

I posted a bug for it.

[Bug]: list index out of range on msg_i in factory.py

[Victor_Lobaco]

The same happens here with ollama 0.6.6 and 0.6.7. I’m using crewai==0.118.0 and crewai-tools==0.43.0.

ERROR:root:LiteLLM call failed: litellm.APIConnectionError: list index out of range
Traceback (most recent call last):
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\main.py”, line 2870, in completion
response = base_llm_http_handler.completion(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\custom_httpx\llm_http_handler.py”, line 269, in completion
data = provider_config.transform_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\ollama\completion\transformation.py”, line 322, in transform_request
modified_prompt = ollama_pt(model=model, messages=messages)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\litellm_core_utils\prompt_templates\factory.py”, line 229, in ollama_pt
tool_calls = messages[msg_i].get(“tool_calls”)
~~~~~~~~^^^^^^^
IndexError: list index out of range

Error during LLM call: litellm.APIConnectionError: list index out of range
Traceback (most recent call last):
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\main.py”, line 2870, in completion
response = base_llm_http_handler.completion(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\custom_httpx\llm_http_handler.py”, line 269, in completion
data = provider_config.transform_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\ollama\completion\transformation.py”, line 322, in transform_request
modified_prompt = ollama_pt(model=model, messages=messages)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\litellm_core_utils\prompt_templates\factory.py”, line 229, in ollama_pt
tool_calls = messages[msg_i].get(“tool_calls”)
~~~~~~~~^^^^^^^
IndexError: list index out of range

An unknown error occurred. Please check the details below.
Error details: litellm.APIConnectionError: list index out of range
Traceback (most recent call last):
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\main.py”, line 2870, in completion
response = base_llm_http_handler.completion(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\custom_httpx\llm_http_handler.py”, line 269, in completion
data = provider_config.transform_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\llms\ollama\completion\transformation.py”, line 322, in transform_request
modified_prompt = ollama_pt(model=model, messages=messages)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File “d:\git\AI_Agents.venv\Lib\site-packages\litellm\litellm_core_utils\prompt_templates\factory.py”, line 229, in ollama_pt
tool_calls = messages[msg_i].get(“tool_calls”)
~~~~~~~~^^^^^^^
IndexError: list index out of range

[Julien]

Same here on mac os 15.4.1
crewai==0.118.0
crewai-tools==0.42.2
ollama==0.4.8
I’m trying to use DOCXSearchTool

Let me know if u found a fix

[Matthew]

Hey @sodoherty - here is what I found on this issue, I also I found a temp solution.. maybe ..

New ollama model qwen3:8b

FYI Crewai Community There is a bug with Qwen3 here is the information Bug Report: IndexError in litellm when CrewAI Agent with Tools uses Ollama/Qwen Affected Libraries: crewai, litellm LLM Provider: Ollama Model: qwen3:4b (or specific Qwen 1.5 variant used) Description: When using CrewAI with an agent configured to use an Ollama model (specifically tested with qwen3) via litellm, an IndexError: list index out of range occurs within litellm’s Ollama prompt templating logic. This er…

[Dom]

I received quite similar error but having tools assigned on task level not agent level. Removing out the tools from task definition helped me finish the crew but I cannot use tools.

LiteLLM 1.67.1
CrewAI 0.118.0
Python 3.12
Windows
Ollama 0.68; gemma3:12b

[Giangi_Sacco]

here is the snippet of the offending part in factory.py

while msg_i < len(messages) and messages[msg_i]["role"] == "assistant":
    assistant_content_str += convert_content_list_to_str(messages[msg_i])
    msg_i += 1
    tool_calls = messages[msg_i].get("tool_calls")

My guess is that is either a bug or when the second condition is verified there was (not anymore) another element in messages, I fixed by adding a check and break right after the msg_i is increased

 while msg_i < len(messages) and messages[msg_i]["role"] == "assistant":
     assistant_content_str += convert_content_list_to_str(messages[msg_i])
     msg_i += 1
     if msg_i == len(messages):
          break
     tool_calls = messages[msg_i].get("tool_calls")

[Abcfsa]

I encountered a similar problem using Ollama 0.6.8,Crewai 0.119.0,model is gwen3:8b
I checked the bugged part in factory.py and found that the “assistant” messages are handled differently with the “user” and “system” messages -
This is the “user” part:

while msg_i < len(messages) and messages[msg_i]["role"] in user_message_types:
    msg_content = messages[msg_i].get("content")
    if msg_content:
        if isinstance(msg_content, list):
            for m in msg_content:
                if m.get("type", "") == "image_url":
                    if isinstance(m["image_url"], str):
                        images.append(m["image_url"])
                    elif isinstance(m["image_url"], dict):
                        images.append(m["image_url"]["url"])
                elif m.get("type", "") == "text":
                    user_content_str += m["text"]
        else:
            # Tool message content will always be a string
            user_content_str += msg_content

     msg_i += 1

Notice where the msg_i +=1 is located at;and this is the “assistant” part:

while msg_i < len(messages) and messages[msg_i]["role"] == "assistant":
    assistant_content_str += convert_content_list_to_str(messages[msg_i])
    msg_i += 1
    
    tool_calls = messages[msg_i].get("tool_calls")
    ollama_tool_calls = []
    if tool_calls:
        for call in tool_calls:
            call_id: str = call["id"]
            function_name: str = call["function"]["name"]
            arguments = json.loads(call["function"]["arguments"])

            ollama_tool_calls.append(
                {
                    "id": call_id,
                    "type": "function",
                    "function": {
                        "name": function_name,
                        "arguments": arguments,
                    },
                }
            )

    if ollama_tool_calls:
        assistant_content_str += (
            f"Tool Calls: {json.dumps(ollama_tool_calls, indent=2)}"
        )
        msg_i += 1

So I changed the “assistant” part so that it aligns with the “user” part:

while msg_i < len(messages) and messages[msg_i]["role"] == "assistant":
    assistant_content_str += convert_content_list_to_str(messages[msg_i])
    # msg_i += 1
   
    tool_calls = messages[msg_i].get("tool_calls")
    ollama_tool_calls = []
    ...
    if ollama_tool_calls:
        assistant_content_str += (
            f"Tool Calls: {json.dumps(ollama_tool_calls, indent=2)}"
        )

    msg_i += 1           

And it worked well then on.This seems to be a bug within the factory.py file,which is a litellm bug to be solved.

[tonykipkemboi]

Current work around also apart from the fix above, is to use crewai version 0.118.0.

We’ve raised the issue with LiteLLM and there seems to be a PR to fix this as well on their end waiting to be merged.

[Dom]

hey @tonykipkemboi
I have crewai version: 0.118.0 but the error is occuring. Same goes for this setup:
crewai-0.119.0 litellm-1.68.0 openai-1.75.0

[Matthew]

5.12.2025 - litellm was just updated to version 1.69.2
Is this update we have been waiting for?

[sodoherty]

I’m still getting the same error. Hot fixing the factory file worked until a proper fix.