Pipedream Python steps → n8n Python Code node

n8n runs Python in the same Code node that runs JavaScript, so a Pipedream Python step has a direct home. Two things change: the handler shape (handler(pd) becomes code that reads _items and returns items) and package imports, which are blocked on n8n Cloud and allow-listed on self-hosted n8n. This page covers both and lists what doesn't translate.

Before and after

Pipedream (Python docs, Python 3.12):

def handler(pd: "pipedream"):
    body = pd.steps["trigger"]["event"]["body"]
    if not body.get("email"):
        return pd.flow.exit("no email")
    return {"email": body["email"].lower()}

n8n Code node, language Python, mode "Run Once for All Items":

out = []
for item in _items:
    body = item["json"]["body"]
    if body.get("email"):
        out.append({"json": {"email": body["email"].lower()}})
return out

In n8n 2.x, native Python exposes only _items (all-items mode) and _item (each-item mode) and uses bracket access, not dot access (Code node docs, 2.0 breaking changes). Older examples using _input.all() were written for the removed Pyodide version.

Mapping table

Pipedream Python	n8n Python Code node
def handler(pd):	Code node body
pd.steps["trigger"]["event"]	_items[0]["json"] when the trigger is the previous node
pd.steps["other_step"]["key"]	Not directly; pass data through the previous node or add a JavaScript Code node that collects it
return {...}	return [{"json": {...}}]
pd.flow.exit()	return []
pd.respond({...})	Respond to Webhook node
os.environ["KEY"]	A credential on the node that needs it
pd.inputs["data_store"]	Data Table node (no static data in Python)
import requests	HTTP Request node
Auto-installed packages	Cloud: none. Self-hosted: installed in the runner image and allow-listed

Because native Python sees only its input items, steps that read several earlier steps need restructuring. A JavaScript Code node can still use $('Node name').

Imports

Pipedream installed packages on import automatically, with a magic comment for name mismatches (import mappings) and version pins such as # pipedream add-package pandas==2.0.0 (Python docs).

In n8n:

• n8n Cloud: "the Python option for the Code node doesn't allow users to import any Python libraries" (Code node docs). Not even json or datetime.
• Self-hosted: Python needs task runners in external mode. Install packages in an image built on n8nio/runners (uv pip install pandas in /opt/runners/task-runner-python) and allow-list them in n8n-task-runners.json under N8N_RUNNERS_EXTERNAL_ALLOW (third-party) and N8N_RUNNERS_STDLIB_ALLOW (standard library). See the task runner docs. By default, imports made inside a package are checked against the allow-list too (N8N_RUNNERS_ALLOW_TRANSITIVE_IMPORTS is false), so list the package's own dependencies as well, or turn transitive imports on.

The runner also blocks some built-ins by default (N8N_RUNNERS_BUILTINS_DENY): eval, exec, open, getattr, setattr, type, dir, globals, __build_class__ and others. Blocking __build_class__ means class definitions fail, so ported code that defines classes needs a rewrite.

Data stores in Python steps

Pipedream Python steps could use data stores through pd.inputs["data_store"] (docs). n8n's workflow static data is JavaScript only, and Data tables can't be read from a Code node. Put a Data Table node before the Python node (to read) or after it (to write). Our converter flags these steps red for this reason. More: Pipedream data stores to n8n.

Should you rewrite in JavaScript?

On n8n Cloud, a Python step that needs any import has to be rewritten anyway, either in plain Python without imports or in JavaScript with HTTP Request nodes doing the API calls. On self-hosted n8n, keep Python if the step depends on libraries like pandas; the runner setup is a one-time job.

Next steps

• Check a Python step for free: code step checker
• See all Python steps in your export: free report
• JavaScript steps: code steps to the n8n Code node
• Full guide: migrate Pipedream to n8n

Not affiliated with Pipedream or n8n.