Sometimes I got it, why? llama_index.core.workflow.errors.WorkflowRuntimeError: Error in step 'run_agent_step': list index out of range

[XinyueZ]

If I used the workflow multiagent stuff, sometimes worked very well, sometimes I got this error:

llama_index.core.workflow.errors.WorkflowRuntimeError: Error in step 'run_agent_step': list index out of range

why? what is the reason?

[dosubot[bot]]

The error "list index out of range" in the run_agent_step method is likely due to the ev.input list being empty or not properly initialized. This can happen if the input list is not correctly populated before the method is called. To resolve this, ensure that the input_messages are properly set up in the init_run step. This involves checking that either user_msg or chat_history is provided and correctly processed to populate input_messages ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[logan-markewich]

You'd need to give the complete traceback, impossible to know without more details

[XinyueZ]

You'd need to give the complete traceback, impossible to know without more details

It's hard to say, bcause every situation is different when an error occurs.

But I want to know under what circumstances this error occurs:

Error: Error in step 'run_agent_step': list index out of range

Code: https://github.com/XinyueZ/chat-your-doc/blob/master/advanced/li_BI_agent_workflow.py

Idea:

Trace:

Traceback (most recent call last):
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/workflow/workflow.py", line 247, in _task
    new_ev = await instrumented_step(**kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/instrumentation/dispatcher.py", line 367, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/agent/workflow/multi_agent_workflow.py", line 290, in run_agent_step
    agent_output = await agent.take_step(
                   ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/agent/workflow/function_agent.py", line 43, in take_step
    async for r in response:
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/llms/callbacks.py", line 88, in wrapped_gen
    async for x in f_return_val:
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/llms/gemini/base.py", line 321, in gen
    content_delta = top_candidate.content.parts[0].text
                    ~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/proto/marshal/collections/repeated.py", line 130, in __getitem__
    return self._marshal.to_python(self._pb_type, self.pb[key])
                                                  ~~~~~~~^^^^^
IndexError: list index out of range

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/bbq/dev/AgenticLlama/code/worked/li_BI_agent_workflow.py", line 411, in <module>
    asyncio.run(main())
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/nest_asyncio.py", line 30, in run
    return loop.run_until_complete(task)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/nest_asyncio.py", line 98, in run_until_complete
    return f.result()
           ^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/asyncio/futures.py", line 203, in result
    raise self._exception.with_traceback(self._exception_tb)
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/asyncio/tasks.py", line 306, in __step_run_and_handle_result
    result = coro.throw(exc)
             ^^^^^^^^^^^^^^^
  File "/home/bbq/dev/AgenticLlama/code/worked/li_BI_agent_workflow.py", line 384, in main
    await agent_workflow.run(
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/asyncio/tasks.py", line 375, in __wakeup
    future.result()
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/workflow/workflow.py", line 445, in _run_workflow
    raise exception_raised
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/asyncio/tasks.py", line 304, in __step_run_and_handle_result
    result = coro.send(None)
             ^^^^^^^^^^^^^^^
  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/core/workflow/workflow.py", line 254, in _task
    raise WorkflowRuntimeError(
llama_index.core.workflow.errors.WorkflowRuntimeError: Error in step 'run_agent_step': list index out of range

[XinyueZ]

By the way, the Gemini 1.5 flash won't have this issue, currently this problem occurs when I use the Gemini 2.0 flash.

[logan-markewich]

So, the source of the error is in the gemini llm class, specifically

  File "/home/bbq/miniconda3/envs/agentic-llama/lib/python3.12/site-packages/llama_index/llms/gemini/base.py", line 321, in gen
    content_delta = top_candidate.content.parts[0].text

Which is weird honestly 🤔 I guess the source code needs to check the length of parts before accessing it. I guess gemini can return messages with no parts?

[XinyueZ]

@logan-markewich Yes, I have the same opinion; it seems that LlamaIndex should fix this with a length check.