Great question. In Prefect 3.x, the “result” lives in result storage and the state just points to it via a DataDocument. In your case, the result was persisted locally at

/root/.prefect/storage/...

with a JSON serializer. Here’s how to retrieve it: CLI - There isn’t a CLI command that resolves and prints the result payload.

prefect flow-run inspect <FLOW_RUN_ID>

will show you the flow run and state metadata (including the storage key), but it won’t load the result content for you. - You can see the help here: CLI: flow-run commands - Command you can run:

prefect flow-run inspect 78b4288d-a6a1-493f-b578-b5d257e47a64

- If the storage key points to a local file you can access (e.g., same machine/container), you can read it yourself (see “Direct file read” below). Python (recommended) - Given a flow run id, Prefect provides a helper to resolve the state’s result:

import asyncio
from prefect.client.orchestration import get_client
from prefect.states import get_state_result

async def fetch_result(flow_run_id: str):
    async with get_client() as client:
        flow_run = await client.read_flow_run(flow_run_id)
    # Resolve and return the Python object stored as the flow's result
    return get_state_result(flow_run.state)

if __name__ == "__main__":
    result = asyncio.run(fetch_result("78b4288d-a6a1-493f-b578-b5d257e47a64"))
    print("Flow result:", result)

- If you need a task result instead, use

read_task_run(task_run_id)

and pass that state to

get_state_result

the same way. Direct file read (when you’re on the same host/container as the worker) - Since your state points to a local storage key and uses the JSON serializer, you can read and decode it directly:

import json
from prefect.serializers import prefect_json_object_decoder

path = "/root/.prefect/storage/5419026f36f948c396a057500d24261c"
with open(path, "rb") as f:
    value = json.loads(f.read(), object_hook=prefect_json_object_decoder)

print("Flow result:", value)

Notes and gotchas - Access to the file path:

/root/.prefect/storage/...

is the worker’s local filesystem path. If you run the retrieval on a different machine/container, you won’t be able to read it unless you use shared/remote result storage (e.g., S3/GCS). The Python

get_state_result

helper will read from whatever storage the state references (local, S3, etc.), provided your environment has access and credentials. - If the result has been cleaned up or storage is inaccessible,

get_state_result

will raise an error. - Docs and references: - Results API (result storage/serializers) - prefect.states.get_state_result source - CLI: flow-run - REST: read flow run state If you share where you’re running the retrieval (same node vs. remote) or if you’re using remote result storage, I can tailor the example (e.g., S3/GCS) and ensure the right dependencies are in place.