Code editor

Run Python 3.13 inside your automation and pass whatever your script prints to downstream Steps.

At a glanceVerified in app · Oct 29, 2026

Use it to: Run calculations, data reshaping, or signature checks in Python 3.13
Find it: Tools → Data → Dev tools → Code editor (t then c)
Cost: 1 Step credit per execution
Limits: Python 3.13 (stdlib, numpy, pandas) · Isolated sandbox (no network)
Output: {{ N.result.output }} (last line), {{ N.result.executionOutput }}, {{ N.result.stderr }}
Instead: Inline pipes ({{ 1.email | trim | lower }}) or Change Fields

Runs Python 3.13 with the standard library, numpy, and pandas. The editor checks syntax as you type and flags unclosed brackets, indentation errors, and bare return statements.

Setting it up

Add the step

Press t then c, or open Tools → Data → Dev tools → Code editor.

Write your Python

Reference earlier steps with unquoted {{ }} placeholders on their own.

Print what the next step needs

Print your result with json.dumps(). The step ignores return values.

Run the automation to see the result

Open the step’s Executions tab to inspect the parsed output.

Reading data from earlier steps

Each {{ }} reference arrives as a typed Python value: JSON objects become dict, arrays become list, and true / false / null become True / False / None. Apostrophes (such as O'Brien) need no escaping.

Valid syntaxStandalone Value Assignment

Place {{ }} unquoted on the right side of an assignment.
Use {{ 3.$full_result }} to read a whole Step output as one dict.
Applies to Step tokens, {{ $var.KEY }}, and {{ $secret.KEY }}.

Rejected syntaxInline Inside Quotes or F-Strings

Never place {{ }} inside quotes: print("Hello {{ 2.name }}") fails.
Assign the token to a Python variable first, then format it.
Standard Python doubled braces in f-strings work normally.

main.py


customer_name = {{ 2.customer_name }}
order_total = {{ 2.total }}
products = {{ 3.$full_result }}
 
# Rejected: print("Hello {{ 2.customer_name }}")
print(f"Hello {customer_name}")

See Variable Reference Syntax.

Available libraries

The sandbox includes Python 3.13’s standard library plus pinned numpy and pandas. Ask support before relying on unlisted packages.

Data libraries (numpy & pandas)

Library	Version	Best For
numpy	`2.5.1`	Numeric arrays and vector math
pandas	`2.3.3`	Tables, grouping, joins, and aggregation

main.py


import numpy as np
import pandas as pd

Task	Module
JSON	`json`
Dates and time zones	`datetime`, `zoneinfo`
Regex	`re`
CSV strings	`csv` with `io.StringIO`
Hashing and HMAC	`hashlib`, `hmac`
Base64 and URLs	`base64`, `urllib.parse`
Grouping and counts	`collections` (`Counter`, `defaultdict`)
Statistics	`statistics`
Exact decimals	`decimal`
IDs and tokens	`uuid`, `secrets`
Iterables	`itertools`, `functools`
Archives	`gzip`, `zipfile`, `tarfile`

What it passes on

A return inside a function does not publish output, and a top-level return is a syntax error. Print your final dictionary on the last line of stdout with json.dumps():

main.py


import json
 
print(json.dumps({
    "final_total": final_total,
    "discount_applied": discount,
}))

Print valid JSON, not Python objects. print(my_dict) outputs single quotes, True, and None. Always serialize with json.dumps().

Reference	Returns	Notes
`{{ 4.result.output.final_total }}`	Parsed JSON from last line	Preferred path. Aliases `{{ 4.final_total }}` and `{{ 4.output.final_total }}` also resolve here.
`{{ 4.result.executionOutput }}`	Raw stdout and stderr	Includes debug `print()` lines up to stream limits.
`{{ 4.result.stderr }}`	Standard error text	Last 20,000 characters of `stderr`, kept separate from `stdout` and never parsed as JSON (omitted when empty).
`{{ 4.$full_result }}`	Full execution wrapper	Returns the `result` wrapper object.
`{{ 4.item.final_total }}` / `{{ 4.results.0.final_total }}`	Looping Code Step output	Read the aligned item (`item`) or index (`results.0`) (loop syntax).

Limits

Sandbox limits

Limit	Value & Behavior
Run time	150 seconds default. Exceeding the timeout stops the script and fails the Step.
Output kept	Up to 100,000 characters of stdout and 20,000 of stderr (256 KiB per-stream cap). Truncating the final stdout line fails with `CODE_OUTPUT_TOO_LARGE`; non-JSON final lines omit `result.output` with a warning.
Memory	Bounded by sandbox capacity. Process large datasets in batches.
Network	None. Opening sockets (`AF_INET` / `AF_INET6`) fails with `PermissionError`.
Concurrency	Isolated workers. Queued bursts wait at `0` Step credits for up to 15 minutes.

What to use instead

HTTP Request (Fetch)Code editor (Transform)HTTP Request (Send)

Python Habit	Sandbox Result	Use Instead
`import requests`	Network blocked	HTTP Request before and after this Step
`smtplib` / DB drivers	No connection	Email or database Steps in Apps
`open("out.csv", "w")`	`Read-only file system`	Write scratch files to `/tmp`, then print JSON or pass text to Drive/S3
Reading a prior run’s file	`FileNotFoundError`	Persist state in Data Store or an app
`input()`	`EOFError`	Reference `{{ 2.field }}` or use Human Review

When a run fails

An uncaught exception fails the Step (rather than returning result.success: false) and stops the branch unless Continue or an Error Path is set (Error Handling). Guard optional keys and validate outputs with Conditions:

main.py


payload = {{ 2.$full_result }}
value = payload.get("response", {}).get("data", {}).get("value", "default")

Examples to copy

Work out a total

main.py


import json
 
order_total = {{ 2.total }}
 
discount = order_total * 0.1 if order_total > 100 else 0
final_total = order_total - discount
 
print(json.dumps({
    "final_total": round(final_total, 2),
    "discount_applied": round(discount, 2),
}))

Score and rank a list

main.py


import json
 
products = {{ 3.$full_result }}
 
scored = []
for product in products:
    score = (product["rating"] * 0.6) + (product["reviews"] * 0.004)
    scored.append({
        "name": product["name"],
        "score": round(score, 2),
        "tier": "premium" if score > 4.0 else "standard",
    })
 
scored.sort(key=lambda p: p["score"], reverse=True)
 
print(json.dumps({"ranked_products": scored}))

Group and summarise with pandas

main.py


import json
import pandas as pd
 
orders = pd.DataFrame({{ 3.$full_result }})
 
by_region = (
    orders.groupby("region")["amount"]
    .agg(["sum", "count"])
    .round(2)
    .reset_index()
)
 
print(json.dumps(by_region.to_dict(orient="records")))

Each example prints JSON on the final line, so downstream Steps read {{ 4.result.output }} directly without Parse JSON.

Verify it works

Run: Open Test & Debug on the Code editor Step and click Run step
You should see: The Output tab shows result.success: true and your printed JSON object parsed under result.output.

What’s Next?

👉 Parse JSON →: expose fields when another Step returns JSON as text.
Helper Functions: use a ready-made conversion instead of maintaining code.
AI Data Transform: reshape unstructured content without code.

Code editor

Setting it up

Add the step

Write your Python

Print what the next step needs

Run the automation to see the result

Reading data from earlier steps

Available libraries

Data libraries (numpy & pandas)

Standard library modules

What it passes on

Limits

Sandbox limits

What to use instead

When a run fails

Examples to copy

Work out a total

Score and rank a list

Group and summarise with pandas

What’s Next?