n8n Audit Kit - client-ready workflow reliability reports

Most n8n work is sold as building. The expensive part is what happens after: the run that dies quietly, the inquiry that gets marked read and never recorded, the retry that replies twice, the handover note that promises something the node cannot do.

This kit is what I use to check workflows before handing them to a client.

WHAT YOU GET

1. n8nwf_lint.py - 12 static checks on an exported workflow. Text or JSON output, exit code for CI.

2. n8n_audit_report.py - turns the same checks into a client-ready HTML report. Your name, client name, print to PDF. Findings never split across pages.

3. A real sample report generated from a 14-node inbound-triage workflow, so you can see the output before you buy.

4. Two ready-made images for your own listing or README.

THE 12 CHECKS

high no-error-workflow failures are silent

high imap-read-before-log mail marked read before it is recorded - inquiries can vanish

high confidence-defaults-high missing confidence treated as certain

high threading-claim-false docs promise In-Reply-To that the Send Email node cannot set

high autosend-hardcoded-on auto-send hardcoded, human gate removed

medium http-no-retry a transient 502 kills the run

medium webhook-open webhook accepts unauthenticated requests

medium no-idempotency a retried trigger is processed twice

medium no-send-log sends mail but records nothing

medium empty-draft-sendable an empty draft can reach the send step

low trigger-doc-mismatch node name contradicts its schedule

low http-no-timeout a hung HTTP call blocks the queue

WHO IT IS FOR

Freelancers and agencies who get asked can you check our workflow, and want to hand over something better than a list of guesses. Anyone who wants a workflow checked in CI before it ships.

HOW TO USE IT

python n8n_audit_report.py your-workflow.json -o report.html --title "Customer support triage" --client "Acme Ltd" --author "Your Name"

Open report.html, Ctrl+P, save as PDF. That is the deliverable.

WHAT IT IS NOT

Static heuristics on the workflow description, not a behavioural test. It does not execute anything, does not see runtime data, and will miss things a human reviewer would catch in a live system. It catches the failure modes visible in the graph - which in my experience are most of the expensive ones.

REQUIREMENTS

Python 3.8+. No dependencies. MIT licence - use it on client work and put your name on the report.

If a check is wrong for your case, email me and I will look at it: abd315222@gmail.com