GoogleSerperAPIWrapper gives error " Input should be a valid string [type=string_type, input_value={'description': 'Dr. Mihi...n field', 'type': 'str'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.10/v/string_type."

[Ansumanbhujabal]

Description

While using GoogleSerperAPIWrapper as a tool in CrewAi Multi agents system , frequently getting this error these days . Which generally makes the agent slow .

I encountered an error while trying to use the tool. This was the error: Arguments validation failed: 1 validation error for Googlesearchtool
search_query
  Input should be a valid string [type=string_type, input_value={'description': 'Dr. Mihi...n field', 'type': 'str'}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.10/v/string_type.
 Tool GoogleSearchTool accepts these inputs: Tool Name: GoogleSearchTool
Tool Arguments: {'search_query': {'description': None, 'type': 'str'}}
Tool Description: Performs a search using the GoogleSearchTool.

Steps to Reproduce

Sample Serper use

search = GoogleSerperAPIWrapper()
@tool("GoogleSearchTool")
def Google_search_tool(search_query: str):
    """Performs a search using the GoogleSearchTool."""
    return search().run(search_query)
## Defining Agents

famous_personality_reviewer_agent = Agent(role=
                          "Senior Content Validator and fact checker",
                          goal="Validate the content with facts and get the most correct content", 
                          backstory=
                          """
                          You are a Senior Content Validator and fact checker  who has been assigned to validate the task 
                          for the {career_name} field if {famous_personalities_name} You will give best fact based 
                          """,
                          llm=llm,
                          tools = [Google_search_tool],
                          verbose=True)

inputs = {
        "career_name": "Automobile Engineer",
       "famous_personalities_name": "Dr. Mihir Shah \nRajendra Singh \nAnupam Mishra \nAyyappa Masagi "
               }

Error is

I encountered an error while trying to use the tool. This was the error: Arguments validation failed: 1 validation error for Googlesearchtool
search_query
  Input should be a valid string [type=string_type, input_value={'description': 'Dr. Mihi...n field', 'type': 'str'}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.10/v/string_type.
 Tool GoogleSearchTool accepts these inputs: Tool Name: GoogleSearchTool
Tool Arguments: {'search_query': {'description': None, 'type': 'str'}}
Tool Description: Performs a search using the GoogleSearchTool.

Dependecies

agentneo==1.2.3
cachetools==5.5.0
certifi==2024.12.14
crewai==0.95.0
crewai-tools==0.32.0
GitPython==3.1.44
google-api-core==2.24.0
google-auth==2.37.0
google-cloud-aiplatform==1.77.0
google-cloud-bigquery==3.27.0
google-cloud-core==2.4.1
google-cloud-resource-manager==1.14.0
google-cloud-storage==2.19.0
google-crc32c==1.6.0
google-resumable-media==2.7.2
googleapis-common-protos==1.66.0
groq==0.13.1
grpc-google-iam-v1==0.14.0
langchain==0.3.17
langchain-cohere==0.3.4
langchain-community==0.3.16
langchain-core==0.3.33
langchain-exa==0.2.1
langchain-experimental==0.3.4
langchain-groq==0.2.2
langchain-openai==0.2.14
langchain-pinecone==0.2.0
langchain-text-splitters==0.3.5
pydantic==2.10.4
pydantic-settings==2.7.1
pydantic_core==2.27.2
scrapegraph_py==1.10.0
selenium==4.27.1
serpapi==0.1.5
setuptools==75.8.0

Expected behavior

This error should not be coming.

Screenshots/Code snippets

Operating System

Ubuntu 20.04

Python Version

3.10

crewAI Version

0.95.0

crewAI Tools Version

0.32.0

Virtual Environment

Venv

Evidence

Possible Solution

None

Additional context

None

[Vidit-Ostwal]

There should be somer error with how you are passing the arguement to the tool, can you share the entire code here once?

[Ansumanbhujabal]

There should be somer error with how you are passing the arguement to the tool, can you share the entire code here once?

Sure , the code simply contains a bunch of more agents like this and their specific tasks.

famous_personality_reviewer_task = Task(description=
                        """
                          Your task is to validate and fact-check .             
                          for the {career_name} field if {famous_personalities_name} are correct famous personality name or not.
                          on the ground of 
                          1. The persons have workied in the {career_name} field directly in the real work.
                          2. THey must predominantly known for this specific {career_name} field , not nay other.
                          3. They have contributed significantly to that field and people consider them as great for their Knowledge,Innovation, Dedication to that specific {career_name} field .
                          4. They are not involved in Criminal offences.
                          5. They are not  passive contributors like investors,industrialists, activists. 
                        """,
                        expected_output=""" Give label "RIGHT" if all names satisfies all grounds , "WRONG" if even one ground is not satisfied """,
                        agent=famous_personality_reviewer_agent)
crew = Crew(agents=[ famous_personality_reviewer_agent], 
            tasks=[famous_personality_reviewer_task], 
            verbose=True)

[Vidit-Ostwal]

@Ansumanbhujabal , I think the problem is because of the \n you are trying to pass

I think even if you remove that and replace with a ',', I think this should work,
Otherway around is you can use """Dr. Mihir Shah \nRajendra Singh \nAnupam Mishra \nAyyappa Masagi""", with triple "" instead of single "",
Try this out once.

[Ansumanbhujabal]

After bumping my head for few days , I found that it was more of a LLM issue for not correctly choosing the format required for tool call . Serper expects a string input , although my input was string , but the llm was passing it as dictionary in tool calling.
refer -https://community.crewai.com/t/tool-calling-not-adhering-to-schema/679/9

I fixed it by explicitly prompting this fact to the agent.

top_companies_reviewer_agent = Agent(role=
                          "Senior Content Validator and fact checker",
                          goal="Validate the content with facts and get the most correct content", 
                          backstory=
                          """
                          You are a Senior Content Validator and fact checker  who has been assigned to validate the task 
                          for the {career_name} field if {top_companies}   are correct companies or not.
                          The tools you have access accept strings as input parameter only.
                          """,
                          llm=llm,
                          tools = [duck_search,Google_search_tool],
                          verbose=True)

Closing this issue since it is not CrewAi specific , feel free to reopen if further problems are faced.