possible hardcoded secrets and token-shaped strings
n8n workflow safety Import slowly
Audit n8n workflow JSON before you activate it.
Workflow packs are useful. Blind imports are not.
This importable n8n workflow scans exported workflow JSON for risky nodes, possible secrets, missing error handling, and AI-to-outbound paths without an obvious approval step.
audit/imported-workflow
$ verdict
do_not_activate_yet
$ rule
import slowly. audit first. activate last.
What it checks
A first-pass audit for workflow JSON.
automatic triggers like webhook, schedule, and cron
email, social, Slack, Telegram, Discord, payment, and database-write nodes
HTTP Request nodes with inline auth or token-looking headers
Code nodes using fetch, axios, eval, or dynamic execution patterns
missing Error Trigger or obvious error workflow
large workflows with no Sticky Notes
AI/LLM nodes connected to outbound actions without an approval step
Why this exists
The n8n ecosystem has a lot of free templates, GitHub repos, and workflow packs. Some are excellent. Some are old, undocumented, or wired to actions you should inspect before activating.
This workflow does the boring first pass: parse the JSON, scan for risky patterns, and return a verdict before you trust the import.
What it returns
-
risk_score: 0-100 score based on finding severity. -
verdict: safe_to_inspect, needs_manual_review, or do_not_activate_yet. -
findings: node-level issues and evidence. -
suggested_fixes: plain-language next steps before activation.
What it is not
This is not a complete security scanner. It will miss obfuscated secrets, custom-node behavior, and runtime-only risks. It is a first-pass audit before activation, not a replacement for reading the workflow.
The rule
Claude/Codex builds. n8n runs. Human approves. For workflows you did not write yourself: import slowly, audit first, activate last.
Use the AI agent workflow safety checklist when the workflow touches publishing, customers, money, production, or private data.
FAQ
What is the n8n Workflow JSON Auditor?
It is an importable n8n workflow that scans exported workflow JSON before activation and flags risky triggers, possible secrets, outbound action nodes, missing error handling, and AI-to-action paths without an approval step.
Is this a complete n8n security scanner?
No. It is a first-pass workflow audit. It helps you slow down before activating an imported workflow, but you still need to inspect credentials, custom nodes, runtime behavior, and business logic yourself.
How do I use the auditor?
Download the JSON, import it into n8n, open the paste node, paste the workflow JSON you want to inspect, run the workflow manually, then review the verdict and suggested fixes before activating the imported workflow.
What does do_not_activate_yet mean?
It means the workflow contains enough risk signals that you should not activate it blindly. Read the findings, remove secrets, inspect risky nodes, add error handling, and add human approval before public, customer-facing, financial, or production actions.