Great question — and just to clarify, do you mean: - the machine where the flow run is actually executing (the worker/infrastructure), or - the machine that submitted/triggered the run (where you ran

prefect deployment run

or used the client)? Here’s how it works in Prefect 3.x: What you can get from inside a running flow - Work pool and queue: available on the current FlowRun - Worker name: recorded as a label on the FlowRun - Hostname of the executing machine: use standard Python (not a Prefect field) - Who triggered the run: available via the FlowRun’s created_by (user or token), but Prefect does not record the submitter’s machine/hostname Example (run inside your flow):

from prefect import flow
from prefect.runtime import flow_run
from prefect.context import FlowRunContext
import socket

@flow
def inspect_runtime():
    ctx = FlowRunContext.get()
    fr = ctx.flow_run if ctx else None

    # Basic flow run info
    print("Flow run ID:", flow_run.id)
    print("Flow run name:", flow_run.name)

    if fr:
        # Worker / pool / queue
        print("Work pool:", fr.work_pool_name)
        print("Work queue:", fr.work_queue_name)

        # Worker name recorded by the worker as a label
        worker_name = (fr.labels or {}).get("<http://prefect.io/worker-name|prefect.io/worker-name>")
        print("Worker name:", worker_name)

        # Who created this run (user/token), not their machine
        creator = getattr(fr.created_by, "display_value", None) if fr.created_by else None
        print("Created by:", creator)

        # Infrastructure/job details if set
        print("Job variables:", fr.job_variables)

    # Actual executing machine identity (OS-level)
    print("Hostname:", socket.gethostname())

Key points - Executing machine: You can infer the worker via labels and get the OS hostname with Python. Prefect does not expose the executing host’s name directly as a Prefect field. - Submitting/triggering machine: Prefect records who created the run (

created_by

) but does not record the submitter’s machine/hostname. Helpful docs - Runtime context overview: Runtime context concepts - Access runtime info: How to access runtime info - FlowRun model (fields like work_pool_name, labels, job_variables, created_by): FlowRun API reference - Context API (FlowRunContext/TaskRunContext): Context API - Work pools and workers concepts: Work pools and Workers If you can confirm which you need (executing machine vs. submitting machine), I can tailor a snippet to your setup (e.g., Process, Docker, or Kubernetes).

Got it — you want info about the machine that submitted/triggered the run. Short answer - Prefect does not record the submitter’s machine/hostname/IP. - Prefect does record who triggered the run via

created_by

(user or token). - If you need submitter machine info, pass it at submission time (as a parameter, tag, or job variable) and read it during the run. What you can get today (who triggered it) Inside your flow:

from prefect.context import FlowRunContext

ctx = FlowRunContext.get()
creator = getattr(ctx.flow_run.created_by, "display_value", None) if ctx and ctx.flow_run.created_by else None
print("Created by:", creator)  # e.g., a user or service token

How to capture the submitter machine going forward Option A — CLI - Pass it as a parameter or job variable when you run the deployment. - Values are parsed as JSON, so strings must be quoted. Examples:

# Parameter (flow must accept the parameter, e.g., submitter_host: str | None = None)
prefect deployment run <flow>/<deployment> --param submitter_host='"'"$(hostname)"'"'

# Job variable (available at runtime via runtime.flow_run.job_variables)
prefect deployment run <flow>/<deployment> --job-variable submitter_host='"'"$(hostname)"'"'

Read it in your flow:

from prefect import flow
from prefect.runtime import flow_run

@flow
def my_flow(submitter_host: str | None = None):
    host_from_param = submitter_host
    host_from_job_vars = (flow_run.job_variables or {}).get("submitter_host")
    print("Submitter host (param):", host_from_param)
    print("Submitter host (job var):", host_from_job_vars)

Option B — Python client Attach submitter info when creating the run.

import socket, datetime, asyncio
from prefect import get_client

async def trigger(deployment_id):
    submitter = {
        "host": socket.gethostname(),
        "time": datetime.datetime.utcnow().isoformat()
    }
    async with get_client() as client:
        await client.create_flow_run_from_deployment(
            deployment_id=deployment_id,
            parameters={"submitter": submitter},   # or {"submitter_host": submitter["host"]}
            # tags=["submitted-from:"+submitter["host"]]  # optional if you want to filter by it
        )

Read it in your flow:

from prefect.runtime import flow_run
submitter = (flow_run.parameters or {}).get("submitter")

Notes and links - This is by design: Prefect records the creator (user/token) but not the creator’s machine. - Runtime context overview: Runtime context concepts - Access runtime info in flows: How to access runtime info - FlowRun model (includes created_by, job_variables, work_pool_name, etc.): FlowRun API reference - Context API (FlowRunContext): Context API If you tell me how you trigger runs (CLI vs. Python, and your shell/OS), I can give you an exact one-liner or snippet with correct quoting.