How to visualize workflow structure - Prefect

You can visualize the structure of your flow using two methods:

flow.visualize() — renders a PNG via Graphviz and opens it in a new window, or inline in Jupyter notebooks. Requires the Graphviz system package.
flow.generate_mermaid_graph() — returns a Mermaid flowchart TD diagram as a string. No additional dependencies required.

<Warning> **These methods execute code outside of tasks**

Code outside of tasks will run when calling visualize() or generate_mermaid_graph(). To avoid inadvertent execution, place code you don't want to run in tasks. </Warning>

Graphviz output

<Note> **Install Graphviz**

To use visualize(), you must have Graphviz installed. Please follow the Graphviz installation instructions for your platform. </Note>

python

from prefect import flow, task

@task(name="Print Hello")
def print_hello(name):
    msg = f"Hello {name}!"
    print(msg)
    return msg

@task(name="Print Hello Again")
def print_hello_again(name):
    msg = f"Hello {name}!"
    print(msg)
    return msg

@flow(name="Hello Flow")
def hello_world(name="world"):
    message = print_hello(name)
    message2 = print_hello_again(message)

if __name__ == "__main__":
    hello_world.visualize()

Mermaid output

generate_mermaid_graph() returns a Mermaid diagram string. This is useful when you want to embed the diagram in GitHub markdown, Notion, or any other tool that supports Mermaid.

python

from prefect import flow, task

@task(name="Print Hello")
def print_hello(name):
    msg = f"Hello {name}!"
    print(msg)
    return msg

@task(name="Print Hello Again")
def print_hello_again(name):
    msg = f"Hello {name}!"
    print(msg)
    return msg

@flow(name="Hello Flow")
def hello_world(name="world"):
    message = print_hello(name)
    message2 = print_hello_again(message)

if __name__ == "__main__":
    diagram = hello_world.generate_mermaid_graph()
    print(diagram)

Because the diagram is returned as a string, you can also write it to a file or embed it in other output:

python

diagram = hello_world.generate_mermaid_graph()
with open("flow.md", "w") as f:
    f.write(f"```mermaid\n{diagram}\n```")

The returned string looks like:

flowchart TD
    Print_Hello_0["Print Hello-0"]
    Print_Hello_Again_0["Print Hello Again-0"]
    Print_Hello_0 --> Print_Hello_Again_0

Paste the output into mermaid.live or any Mermaid-compatible renderer to view the diagram.

Dynamic flows with mock return values

For workflows with dynamic structure using loops and/or if/else statements, you can provide tasks with mock return values to choose which code paths to visualize.

python

from prefect import flow, task

@task(viz_return_value=[4])
def get_list():
    return [1, 2, 3]

@task
def append_one(n):
    return n.append(6)

@flow
def viz_return_value_tracked():
    l = get_list()
    for num in range(3):
        l.append(5)
        append_one(l)

if __name__ == "__main__":
    viz_return_value_tracked.visualize()