Tools & Workflows · Request to task with n8n

Check first. Then create a task.

Request → task · English edition of 25 September 2026 · German text as of 8 September 2026

The business rule is small: a complete new request produces exactly one proposed task. Missing details and repeated requests each get a result of their own that you can check. Here, a task record is a proposal in the result list, not yet a ticket in your target system.

What you need before importing

A separate n8n test instance or a suitable cloud test area, the workflow and the synthetic sample data. Never import into a workflow that is already in production use. The package needs no email, CRM or AI credentials.

The setup guide and the sample data are in German. The workflow itself runs the same way in any language.

The rule stays visible in the result

  1. Look at the input. Open the supplied request. Its ID is the basis of the duplicate check; the task does not get a new random key on every attempt.
  2. Check required fields. A value made up only of spaces does not count as filled in. Missing or malformed values are not silently filled in.
  3. Handle repeats. Within one batch, only the first valid request with a given ID creates a task. Every further valid request with the same ID is reported as a duplicate, even if its text differs. If needed, compare changes against the source file.
  4. Read the outputs. Check successful tasks, duplicates and errors separately. The number of successful runs alone does not tell you whether a request was lost.

Import and test deliberately

The export contains a manual trigger, the synthetic requests, the validation node and the result formatting. Open a new workflow and import n8n-workflow.json using the file import in the workflow menu. Leave the supplied synthetic inputs unchanged at first. Then follow the version-specific README in the download; it is the exact guide for this export version.

Start the manual test and compare the complete outputs with expected-results.json. The third supplied request repeats the first ID and must end up in the duplicate path. A second full workflow run processes the batch again: the duplicate check only remembers IDs within this batch, not across runs.

Download the cloud test procedure (in German) ↓ · Expected results and error cases

What still separates the demo from real operation

Duplicate protection that stays reliable over time needs a persistent, unique key in the target system. “Look first, then create” can write twice when two requests arrive at the same time. The demo makes no promise about concurrent operation or exactly-once delivery across several systems.

For real tasks, you need to connect the target system, restrict access and be able to reprocess errors after an outage. Decide who reads the error list every day. Only go live once failed writes are visible and a retry does not create a second task.

The measured result

Run on 6 September 2026 with n8n 2.37.10. Each of the seven inputs is assigned to a visible output:

Error reasons: missing email, invalid email syntax and empty subject. The sample addresses are synthetic. No message was sent.

Test report with version and file hashes ↓

n8n Cloud or self-hosted?

With n8n Cloud, n8n runs the infrastructure. If you host it yourself, you take on updates, backup and recovery, among other things. Either way, you have to set up your workflows, access rights and error handling. n8n explains the hosting options ↗.

See the same case in Make →