litellm.exceptions.UnsupportedParamsError with Bedrock on Final LLM Call

[lightme16]

Description:

When using Holmes GPT with an AWS Bedrock model for investigation, an UnsupportedParamsError from litellm can occur. This seems to happen specifically during the final step of the investigation loop within ToolCallingLLM, when Holmes attempts to get a final text response from the LLM after potentially multiple tool calls. I am using it with bedrock/us.anthropic.claude-3-7-sonnet-20250219-v1:0

Traceback:

Running tool kubectl_get_by_kind_in_namespace: kubectl get          tools.py:127
--show-labels -o wide hpa -n otcollector                                        
Traceback (most recent call last):
  File "/Users/ozviriuk/Applications/PyCharm Community Edition.app/Contents/plugins/python-ce/helpers/pydev/pydevd.py", line 1570, in _exec
    pydev_imports.execfile(file, globals, locals)  # execute the script
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/ozviriuk/Applications/PyCharm Community Edition.app/Contents/plugins/python-ce/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/main.py", line 928, in <module>
    run()
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/main.py", line 924, in run
    app()
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/typer/main.py", line 338, in __call__
    raise e
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/typer/main.py", line 321, in __call__
    return get_command(self)(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/click/core.py", line 1161, in __call__
    return self.main(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/typer/core.py", line 728, in main
    return _main(
           ^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/typer/core.py", line 197, in _main
    rv = self.invoke(ctx)
         ^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/click/core.py", line 1697, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/click/core.py", line 1443, in invoke
    return ctx.invoke(self.callback, **ctx.params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/click/core.py", line 788, in invoke
    return __callback(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/typer/main.py", line 703, in wrapper
    return callback(**use_params)
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/main.py", line 305, in ask
    response = ai.prompt_call(system_prompt, prompt, post_processing_prompt)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/core/tool_calling_llm.py", line 120, in prompt_call
    return self.call(
           ^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/sentry_sdk/tracing_utils.py", line 823, in func_with_tracing
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/core/tool_calling_llm.py", line 174, in call
    full_response = self.llm.completion(
                    ^^^^^^^^^^^^^^^^^^^^
  File "/Users/ozviriuk/src/holmesgpt-test/vendor/holmesgpt/holmes/core/llm.py", line 215, in completion
    result = litellm.completion(
             ^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/utils.py", line 1247, in wrapper
    raise e
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/utils.py", line 1125, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/main.py", line 3148, in completion
    raise exception_type(
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/main.py", line 2664, in completion
    response = bedrock_converse_chat_completion.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/llms/bedrock/chat/converse_handler.py", line 377, in completion
    _data = litellm.AmazonConverseConfig()._transform_request(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/llms/bedrock/chat/converse_transformation.py", line 568, in _transform_request
    _data: CommonRequestObject = self._transform_request_helper(
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/venv/holmesgpt/lib/python3.12/site-packages/litellm/llms/bedrock/chat/converse_transformation.py", line 445, in _transform_request_helper
    raise litellm.UnsupportedParamsError(
litellm.exceptions.UnsupportedParamsError: litellm.UnsupportedParamsError: Bedrock doesn't support tool calling without `tools=` param specified. Pass `tools=` param OR set `litellm.modify_params = True` // `litellm_settings::modify_params: True` to add dummy tool to the request.

Should we add more validations for that to handle it better?