Lex alone is not designed to dynamically handle open-ended questions without predefined utterances.
Here's how to capture arbitrary user questions and pass them to your Lambda function using a workaround approach:

1. Create a Catch-All Intent in Lex (e.g., FallbackIntent or CatchAllQuestionIntent).

2. Enable Lambda Fulfillment on that intent.

3. Extract User Input in Lambda and use it to search your knowledge base.

4. Respond Dynamically with the relevant answer.

Step-by-Step Instructions

1. Create a Catch-All Intent

Use this intent to catch anything that doesn't match other specific intents:

intents:
  - name: CatchAllQuestionIntent
    fulfillmentCodeHook:
      enabled: true
    sampleUtterances:
      - "I have a question"
      - "Can you tell me something?"
      - "What is {UserQuestion}"
      - "Tell me about {UserQuestion}"
      - "Explain {UserQuestion}"
    slots:
      - name: UserQuestion
        slotType: AMAZON.SearchQuery
        valueElicitationSetting:
          slotConstraint: Required
          promptSpecification:
            messageGroupsList:
              - message:
                  plainTextMessage:
                    value: "What would you like to know?"
            maxRetries: 2
          slotCaptureSetting: {}

2. Use the AMAZON.SearchQuery Slot Type

This allows the user to type in anything freely and captures their question as-is.

3. Pass to Lambda

In your Lambda function, extract the UserQuestion slot and query your knowledge base:

def lambda_handler(event, context):
    question = event['sessionState']['intent']['slots']['UserQuestion']['value']['interpretedValue']

    # Your logic to search the YAML-based knowledge base
    answer = search_kb(question)

    return {
        "sessionState": {
            "dialogAction": {
                "type": "Close"
            },
            "intent": {
                "name": event['sessionState']['intent']['name'],
                "state": "Fulfilled"
            }
        },
        "messages": [
            {
                "contentType": "PlainText",
                "content": answer
            }
        ]
    }