Python in the n8n Code Node (n8n 2.x): Setup, Syntax and Limits

Python in n8n changed significantly. The first implementation ran Python through Pyodide, a WebAssembly port of CPython. n8n then introduced native Python on task runners, available from version 1.111.0 and stable as of n8n 2. Pyodide is removed in n8n 2.0. If your old Python snippets now fail, this is why. Here is how the current version works.

What changed from Pyodide to native Python

According to the n8n documentation, the main differences are:

Upgrading is a breaking change, so plan to rewrite snippets that used dot access such as item.json.myField.

The syntax that works

Run Once for All Items

out = []
for item in _items:
    data = item["json"]
    out.append({"json": {
        "email": str(data.get("email", "")).strip().lower(),
        "company": data.get("company"),
    }})
return out

Run Once for Each Item

data = _item["json"]
data["full_name"] = f'{data.get("first_name", "")} {data.get("last_name", "")}'.strip()
return _item

The return format is the same as in JavaScript: a list of dictionaries, each with a "json" key holding a dictionary.

Allowing imports on self-hosted n8n

By default native Python blocks all imports, even json or datetime from the standard library. Two variables open them:

Third-party packages must also exist in the runner image (n8nio/runners in the official setup). The documentation describes adding extra dependencies by extending that image. In external mode, set these variables in the runners configuration, not on the main n8n container.

# With N8N_RUNNERS_STDLIB_ALLOW=re,datetime
import re
from datetime import datetime, timedelta
out = []
for item in _items:
    phone = re.sub(r"[^0-9+]", "", str(item["json"].get("phone", "")))
    follow_up = (datetime.utcnow() + timedelta(days=3)).strftime("%Y-%m-%d")
    out.append({"json": {**item["json"], "phone": phone, "follow_up": follow_up}})
return out

n8n Cloud

On n8n Cloud the Python option of the Code node does not allow importing any library, from the standard library or third parties. Plain Python works; import does not. If you need a library on Cloud, use JavaScript (which has crypto and moment), a dedicated n8n node, or call an external function with the HTTP Request node.

Python or JavaScript?

Choose JavaScript when:

• you need n8n helpers such as $('Node name'), $execution, $now or binary data functions, which native Python does not support;
• you are on n8n Cloud and need any library;
• the next person maintaining the workflow does not read Python.

Choose Python when the logic itself is easier in Python (text processing, data science libraries you are allowed to install) and the input arrives through the node's own items.

Worked example: migrate a Pyodide snippet to native Python

Many workflows built in 2023 and 2024 contain snippets like this one, which stopped working when Pyodide was removed:

# Old (Pyodide), no longer supported
for item in _input.all():
    item.json.email = item.json.email.lower().strip()
    item.json.domain = item.json.email.split("@")[1]
return _input.all()

The native version, for "Run Once for All Items":

out = []
for item in _items:
    d = dict(item["json"])
    email = str(d.get("email", "")).strip().lower()
    d["email"] = email
    d["domain"] = email.split("@")[1] if "@" in email else None
    out.append({"json": d})
return out

What changed, line by line:

• _input.all() became _items;
• item.json.email became item["json"]["email"], read safely with .get();
• the item is rebuilt as a dictionary with a "json" key before returning;
• the domain split is guarded, because native Python raises an error on a missing @ just like the old code did, but now you see it instead of a vague failure.

Allowing libraries on self-hosted n8n

Three steps, in this order:

1. 1 Install third-party packages in the runner image (the official runner image is n8nio/runners), by extending it as the n8n documentation describes.
2. 2 Allow them: standard-library modules in N8N_RUNNERS_STDLIB_ALLOW, third-party modules in N8N_RUNNERS_EXTERNAL_ALLOW. Use names, not *, unless the instance is single-user.
3. 3 Restart the runners.

There is also N8N_RUNNERS_ALLOW_TRANSITIVE_IMPORTS: by default, imports made inside an allowed package must be allowed too. Turning it on lets an allowed package load its own dependencies, at the cost of a wider surface.

When native Python is the right tool

Troubleshooting table

Production checklist

• n8n version checked: native Python, not Pyodide, from n8n 2.0.
• Bracket notation everywhere (item["json"]["field"]).
• Imports allowed explicitly and installed in the runner image (self-hosted).
• JavaScript chosen instead when you need n8n helpers or Cloud libraries.

Frequently asked questions

Why did my Python Code node break after upgrading n8n?

n8n 2.0 removed Pyodide. Native Python uses bracket notation (item["json"]["field"]) and only supports _items and _item. Rewrite dot access and remove other _ helpers.

How do I import a Python library in n8n?

On self-hosted n8n, list it in N8N_RUNNERS_STDLIB_ALLOW (standard library) or N8N_RUNNERS_EXTERNAL_ALLOW (third-party), and make sure third-party packages are installed in the runner image. On n8n Cloud, imports are not allowed.

Can I read binary files with Python in n8n?

The getBinaryDataBuffer() helper is not supported in Python. Use JavaScript or the Extract from File node for that step.

Is Python slower than JavaScript in n8n?

The old Pyodide version was noticeably slower. Native Python runs in a real Python process on task runners. For typical workflow volumes either language is fast enough; pick the one your team maintains best.

Is Python faster or slower than JavaScript in n8n?

For typical workflow volumes both are fast. JavaScript has the advantage of n8n's built-in helpers; Python is worth it when the logic or library is genuinely better in Python.

Can I use pandas in n8n?

On self-hosted n8n with native Python, if pandas is installed in the runner image and allowed with N8N_RUNNERS_EXTERNAL_ALLOW. Not on n8n Cloud.

Which n8n version do I need for native Python?

Native Python on task runners is available from n8n 1.111.0 and stable as of n8n 2. Pyodide is removed from n8n 2.0.

Steve, founder of PURIST

Steve founded PURIST, a workflow automation studio working with n8n, Make and Claude AI, and is the point of contact on every project. Prices and features in our articles come from vendor documentation and public pricing pages. How we research and score tools.