Running n8n Workflow JSON via CLI (No UI) on Ubuntu

Questions
, ,
1

Hi everyone,

I’m looking for guidance on executing an existing n8n workflow purely via CLI, without using the n8n UI at all.

Context

  • I have a workflow.json
  • The workflow is already fully defined (nodes, connections, etc.)
  • I want to run this workflow locally on Ubuntu
  • I do NOT want to run via n8n editor UI

Goal
I’m trying to achieve something like:

  • Load workflow JSON
  • Execute it directly from terminal / CLI
  • Pass the credentials via CLI
  • Get logs/output in the terminal

Example

  • Execute a simple workflow that interacts with a GPT model
  • Capture and print the generated text output

Questions

  1. Is there a way to execute a workflow JSON directly using the n8n CLI (e.g., n8n execute or similar)?
  2. Do I need to first import the workflow into n8n’s database, or can it be run as a standalone file?
  3. Is there a recommended approach using:
    • n8n execute
    • n8n run
    • or any headless mode?
  4. How are credentials handled in this scenario (since UI is not being used)?
  5. Any best practices for running workflows in a fully headless / automation-friendly setup?

What I Tried

  • I tried running n8n via Docker, but I’m running into a port issue on 5679 (port conflict / binding error)
    • n8n Task Broker’s port 5679 is already in use. Do you have another instance of n8n running already?

If anyone has experience running n8n workflows in a fully CLI-driven or headless environment, I’d really appreciate your guidance :folded_hands:
Thanks in advance!

2

Hi @Stephanc Welcome!
First i would recommend reading this:

and now for doing that kind of direct execution from CLI, i think first you need to import your flow via:
n8n import:workflow --input=workflow.json
and then once you have got the ID for that, you can execute it using that ID:
n8n execute --id <ID>
and for your port conflict i think you should try:
N8N_RUNNERS_ENABLED=false or also if you want you can just flip and change the port with N8N_RUNNERS_BROKER_PORT.

for credentials i would say they must be pre imported ofc else how you would be able to use that and you cannot pass them via CLI flags or command kind of.

1 Like
3

yeah, the import + execute is solid. main headache is credentials though — since youre headless theres no UI to enter them. if youre on docker, feeding them via env vars + the --credentials flag works pretty smoothly. what kind of credentials are you dealing with?

4

@Anshul_Namdev @Benjamin_Behrens

I tried running n8n with N8N_RUNNERS_ENABLED=false. Below is the Docker command I used:

sudo docker run -d \
 --name n8n \
 --restart unless-stopped \
 -p 5678:5678 \
 -e GENERIC_TIMEZONE="Europe/Warsaw" \
 -e TZ="Europe/Warsaw" \
 -e N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS=true \
 -e N8N_RUNNERS_ENABLED=false \
 -v n8n_data:/home/node/.n8n \
 docker.n8n.io/n8nio/n8n

Here is the workflow JSON that I want to execute via CLI:

sudo docker exec -it n8n n8n export:workflow --all | jq '.[] | {id, name}'

{
  "id": "cs1aJafzGKWZrXH2",
  "name": "openai"
}

sudo docker exec -it n8n n8n execute --id=cs1aJafzGKWZrXH2
n8n Task Broker's port 5679 is already in use. Do you have another instance of n8n running already?

can u help me with the solution. ?

5

the port 5679 conflict happens because the running n8n instance already claimed that port — even with N8N_RUNNERS_ENABLED=false, the CLI execution tries to start its own broker process.

two options:

option 1 — override the broker port for the exec:

sudo docker exec -it -e N8N_RUNNERS_BROKER_PORT=5680 n8n n8n execute --id=cs1aJafzGKWZrXH2

the -e flag lets you pass a different port just for that exec session.

option 2 — use the REST API instead (cleaner for automation):

curl -X POST http://localhost:5678/api/v1/workflows/cs1aJafzGKWZrXH2/run \
  -H "X-N8N-API-KEY: your-api-key"

generate the API key under Settings → n8n API. this avoids port conflicts entirely and works well for scripting. credentials stay in the container’s db, no CLI conflicts.

6

thank you.

I tried with both the options that you gave
for the option 1, i got error

n8n Task Broker ready on 127.0.0.1, port 5680
Failed to start Python task runner in internal mode. because Python 3 is missing from this system. Launching a Python runner in internal mode is intended only for debugging and is not recommended for production. Users are encouraged to deploy in external mode. See: https://docs.n8n.io/hosting/configuration/task-runners/#setting-up-external-mode
Execution was NOT successful. See log message for details.
Execution error:
====================================

and for the Option 2 , after generating the API key and running the curl, it says

{
    "message": "POST method not allowed"
}
7

the python warning is a red herring — your workflow only has JS nodes so that doesnt affect execution.

the real problem with option 1 is a credential mismatch. your workflow JSON references credential id 14uuAWvu47W8hbvx (“OpenAI account”), but that id comes from the original n8n instance. in your fresh docker container, that credential simply doesnt exist — so the execution fails even though the broker started fine on port 5680.

fix: open the new n8n instance in the browser, go to Credentials → create a new OpenAI credential, then update the workflow JSON with the new credential id before importing and running it.

for option 2 — the “POST method not allowed” is expected with a Manual Trigger. that trigger type is UI-only by design and cant be invoked via REST API.

to make the curl approach work, swap the first node for a Webhook node:

  1. replace manualTrigger with a Webhook node in your workflow JSON
  2. activate the workflow in n8n
  3. call the webhook URL directly — no port conflicts, no credential workarounds

thats the cleanest path for any scripted/headless execution anyway.