Issue with TxtSearchTool missing keyword arguments

CrewAI Community Support
1

I think I am correctly initializing TxtSearchTool but i get an error

from crewai_tools import TXTSearchTool

txt_search_tool = TXTSearchTool(
	config={
		'llm': {
			'provider': 'azure_openai',
			'config': {
				'model': os.environ.get("model"),		
			}
		},
		'embedder': {
			'provider': 'azure_openai',
			'config': {
				'model': 'text-embedding-ada-002-2',
				'deployment_name': 'text-embedding-ada-002',
				'api_key': os.environ.get("AZURE_API_KEY"),
			}	
		}
	}

This produces an error when trying to run the task:

# Agent: Senior Scientific Researcher
## Using tool: Search a txt's content
## Tool Input: 
"{\"search_query\": \"happiness\", \"txt\": \"path/to/txt/file"}"
## Tool Output: 

I encountered an error while trying to use the tool. This was the error: APIStatusError.__init__() missing 2 required keyword-only arguments: 'response' and 'body'.
 Tool Search a txt's content accepts these inputs: Tool Name: Search a txt's content
Tool Arguments: {'search_query': {'description': "Mandatory search query you want to use to search the txt's content", 'type': 'str'}, 'txt': {'description': 'Mandatory txt path you want to search', 'type': 'str'}}
Tool Description: A tool that can be used to semantic search a query from a txt's content..
Moving on then.

Seems like the LLM doesn’t understand the tool’s inputs, but “response” and “body” are not obvious parameters from the docs or the tool’s code.

2

After more tinkering, I think this is due to a failed attempt to reach the model’s API.

3

I am having the same issue with the Website WebsiteSearchTool, what was the root cause?

4

It means crewai cannot reach the LLM provider, so either it’s some proxy issue or the api key is wrong or the url to the LLM is wrong