The task cannot see the parameter values

[jcppython]

Bug summary

code demo

from prefect import flow, task
from typing import List, Dict, Optional

@task(name="test_task")
def process_data(
    data: List[Dict],
    threshold: float = 0.5,
    config: Optional[Dict] = None
) -> Dict:
    result = {"processed": len(data), "threshold": threshold}
    if config:
        result.update(config)
    return result

@flow(name="test_flow")
def my_flow(data_source: str, config: Dict = {"mode": "normal"}):
    data = [{"id": i, "value": i/10} for i in range(10)]
    
    result = process_data(data, threshold=0.7, config=config)
    return result

if __name__ == "__main__":
    my_flow(data_source="example_data", config={"mode": "debug"})

Version info

Version:             3.2.15
API version:         0.8.4
Python version:      3.12.9
Git commit:          0f5a3081
Built:               Fri, Mar 28, 2025 3:28 PM
OS/Arch:             linux/x86_64
Profile:             ephemeral
Server type:         server
Pydantic version:    2.10.3

Additional context

No response

[zzstoatzz]

hi @jcppython - this is by design, task run parameter values are currently not stored in the API (which makes tasks faster and oftentimes task inputs are not serializable). if you click on the tooltip for Task Run Inputs you'll get

do you want me to convert this to a discussion?

[jcppython]

Perhaps we could have some asynchronous ways to pull parameters, which can both meet the needs of problem - troubleshooting and avoid affecting performance.

Okay, I'm glad to discuss this issue. @zzstoatzz

[zzstoatzz]

Generally, the way people deal with this is by selectively logging things that they want to see for debugging purposes. I'm not sure what you mean by asynchronous ways to pull parameters. Task run parameter values are simply not stored in the prefect backend. So if you want them to be accessible in the UI, you have to print them or otherwise save them in a way that they can be rendered.