Validation error for TaskOutput

CrewAI Community Support
1

Hello, I’m trying to get structured outputs, but I’m having trouble.

Problem Description

I’m encountering a persistent validation error in my CrewAI implementation when processing agent outputs through Pydantic. The task requires an agent (“Glancy”) to analyze missing requirements and output a JSON object matching this model:

class requirements_model(BaseModel):
    isComplete: str = Field(description="Boolean indicating requirement completeness")
    requirements: str = Field(description="Analysis of missing requirements")

Observed Failure

The task consistently fails with this validation error:

ValidationError: 1 validation error for TaskOutput
json_dict
  Input should be a valid dictionary [type=dict_type, input_value='{...}', input_type=str]

Key Details

  1. Agent Output: Appears as valid JSON string:
# Agent: Interpret Agent
## Final Answer:
I am an Interpret Agent. My precision and attention to detail are unmatched,   
and my ability to understand and interpret complex information is unparalleled.
Based on the provided code and test results, I have determined that the final  
answer should be in JSON format with two keys: "isComplete" and "requirements".
Here is the final answer:

{
  "isComplete": "true",
  "requirements": [
    {
      "name": "Gearbox_fsm_type",
      "type": "enum",
      "values": ["First", "Second", "Third", "Fourth"]
    },
    {
      "name": "input",
      "type": "struct",
      "fields": ["In_speed", "In_Throttle"]
    },
    {
      "name": "output",
      "type": "struct",
      "fields": ["Out_gear"]
    },
    {
      "name": "internal",
      "type": "struct",
      "fields": ["fsm_var"]
    }
  ]
}

I have thoroughly reviewed the provided code and test results to ensure that myfinal answer is accurate and complete. I am confident that this answer meets   
all requirements and will be accepted.
  1. CrewAI Configuration:
#### TASKS .yaml
interpret_missing_requirements:
  description: >
    Your task is to draft a detailed report based on the missing requeriments you found in:
    {missing_requirements}

    Take into account the last generated code:
    {generated_code}

    If you do your BEST WORK, you'll earn a $10,000 commission!
  expected_output: >
   A JSON object with a 2 keys: isComplete and requirements.
  agent: Glancy

#### CREW .py
@task
def interpret_missing_requirements(self) -> Task:
    return Task(
        config=self.tasks_config['interpret_missing_requirements'],
        output_json=requirements_model,
    )

What am I doing erroneously?

Thanks in advance!