Field guide / automation reliability / November 9, 2026

Can a handoff checker beat a free manual review?

A repeatable, deliberately small experiment for n8n operators. Use one completely fictional workflow, compare two methods, record what each method misses, and accept that the free method might win.

Get the free comparison scorecard (CSV) ↓ Run the fictional beta test →

The problem worth measuring

Someone receives an n8n workflow export from a colleague or an automation agency. The JSON opens, but a handoff is not complete: the next operator needs to identify credential references, know which accounts require setup, understand nodes that carry those references, and verify the new environment through authorized execution. Merely seeing credential names is not proof that a credential works.

According to n8n's export/import guidance, workflow exports can include credential names and IDs. An export should therefore be inspected and sanitized before sharing. The official cloud-to-self-hosted guidance also explains why exporting workflows alone does not transfer working credential secrets. These are separate limitations that a text-only review cannot solve.

The Moruneva local handoff map scans exported credential references and produces a planning checklist. It cannot inspect live secrets, validate access, execute the automation, or certify security. We need independent users to tell us whether the planning step actually gets easier.

What you need

  1. The four-node fictional workflow JSON. It contains placeholder, unusable references. Use this exact sample for both methods.
  2. A local text editor or JSON viewer for the manual baseline. You do not need Moruneva for the first pass.
  3. The free browser-local map for the second pass.
  4. The five-column CSV scorecard, opened in any spreadsheet or text editor. No signup or data upload is required.

Strict safety rule: only use the fictional sample. Do not submit actual customer workflow exports, passwords, tokens, personal information, account names, or private contracts. The CSV and the comparison results are yours to keep locally.

Protocol: 10 minutes, one task, two methods

1. State the job before looking at either result

Imagine you are receiving a workflow you did not build. Your goal is to produce a reviewable list of credential-reference locations and unresolved setup questions. You are not trying to prove a production transfer succeeded. Keep this goal identical in both passes.

2. Use an existing free method first

Open the fictional JSON in a text editor or your usual free inspection tool. Locate the credentials objects and write down which node points to each credential type, how many references you found, and the questions you would ask the new owner. Start timing when you begin reading and stop when your list is complete. Do not open Moruneva while doing this pass.

3. Run Moruneva on the same sample

Load the same fictional file (or the built-in equivalent) into the browser-local map. Select Review locally, read the reference summary, and download the anonymous handoff checklist. Start a separate timer immediately before loading and stop only when you have reviewed the output. Record corrections and manual follow-up work. Do not credit the tool for checks it did not perform.

4. Count omissions and disagreements, not just speed

Compare both lists against the fictional JSON. Distinguish a missed credential reference from an unresolved real-world setup question. Record false positives, skipped references, and ambiguous outputs separately. A faster list that misses a critical reference may be worse. If learning the sample on the first pass made the second faster, record that ordering bias.

5. Ask which one you would actually reuse

For a second, unfamiliar fictional export next week, would you choose your existing manual method or this local tool? If you would choose Moruneva, what exact part saved you work? If not, what was redundant or misleading? A preference is useful evidence, but it is not willingness to pay. Do not infer a price from one test.

How to score fairly

ObservationInterpretation
Elapsed timeCompare end-to-end minutes, including checking and correcting the output. One trial is illustrative, not a statistical result.
Credential references foundCheck the node positions and types against the fictional source; fewer entries may signal an omission.
Unresolved setup questionsKeep these visible. Neither method is entitled to claim runtime access or permissions from JSON alone.
Incorrect or confusing entriesCount false positives, duplicate confusion, or wrong node attribution as real costs.
Would use again?Record a reason, including “no”. A useful free tool is allowed to have no paid market.

For additional fairness, repeat with a second fictional workflow on another day and reverse the method order. The provided four-node sample is deliberately tiny; it may not expose bottlenecks seen in larger automations. Never label this exercise a controlled scientific trial or production security audit.

What would change our product decision?

If users consistently finish faster without extra omissions, a local checklist may be worth maintaining. If free text inspection already meets the job, we should leave this as a free experiment instead of selling a duplicate. If both methods leave the same important setup questions unanswered, better prompts or a different product could matter more than a prettier reference list.

The first independent volunteer comparison is still being recruited; we have no published independent result, verified paying customer, or evidence of commercial uplift. This article documents a falsifiable testing method, not an outcome.

Try it and disagree with us

Experienced n8n operators are welcome to test with fictional data. One concrete omission is more valuable than a polite endorsement. You may keep the CSV private, or voluntarily share only non-confidential observations.

See the voluntary 5–10 minute beta → Email a short observation ↗

Further reading: n8n: exporting and importing workflows · n8n: migrating from cloud to self-hosted.