350 finished automations, free — the public folder you can run in a minute

# 350 automations somebody already finished

Almost nobody stalls on *what* to automate. You already know the thing you retype every morning.
You stall an hour and a half into a blank canvas, on the node that will not authenticate, with
nothing to show for the evening.

awesome-n8n-templates is a public GitHub repository holding 350 finished n8n workflow files.
Not a course, not a list of ideas — 350 .json exports of machines that already ran for somebody
before they reached you. 25,162 stars and 6,433 forks as of 2026-09-05, and still accepting contributions.

## What is actually in there

Nineteen folders, named after the apps you already open. These counts are read off the
repository's own git tree, not off anybody's caption:

Plus one stray .json at the repository root, which is where the 350th file comes from.

1. Open the folder named after the app you already do the chore in.
2. Take exactly one .json out of it.
3. In n8n: import the workflow file.
4. Connect the credentials it asks for.
5. Run it once, on a real input.

That is the whole loop, and it genuinely does take about a minute of clicking. The reason it works
is that a workflow export carries the full node graph — every step, every mapping, every parsed
field — so there is nothing left to design.

The repository brings no accounts and no keys. That is obvious when you say it out loud and it
is still where most people stall, because it does not stall you at import time — it stalls you at
run time, one credential at a time, after you have already imported six workflows.

There is a second, quieter thing. These exports were made from somebody's live n8n. n8n does not
export secrets, but it does export credential *names*, and those names are frequently real
addresses. Open Auto-label incoming Gmail messages with AI nodes.json and you can read that it
was wired to Gmail (workfloowstutorial@gmail.com) and OpenAI (workfloows@gmail.com). No secret
leaks — but it tells you plainly that you are holding a snapshot of a working setup, not a blank
template. It is worth reading before you trust it with your inbox.

credcheck.py in the kit reads a workflow file before you import it and prints the credential
types it will demand, every node type it touches, whether it carries a webhook, and whether it is
marked active. Offline, read-only, standard library only:

python3 credcheck.py Gmail_and_Email_Automation/some-workflow.json

nodes        : 19
active flag  : False
YOU MUST CONNECT:
  - gmailOAuth2   (the export was wired to: Gmail (workfloowstutorial@gmail.com))
  - openAiApi     (the export was wired to: OpenAI (workfloows@gmail.com))
touches      : aggregate, chainLlm, gmailx3, gmailTrigger, lmChatOpenAi, merge, ...

Two connected accounts, not one. Knowing that before the import is the difference between a
workflow that runs today and a workflow that sits in your instance, red, until you forget why you
downloaded it.

Not the biggest one. OpenAI_and_LLMs/ has 92 files and is the wrong place to start, because "use
a model" is a capability, not a chore — you cannot tell whether it worked.

Name the one thing you retype every morning, then open the folder named after the app you retype
it into. Import one. Connect its credentials. Run it once, end to end. Only then open a second.

A shelf of 350 free machines does not fail you by being wrong. It fails you the evening you drag
half of it into your instance and connect none of it.

- Repository: <https://github.com/enescingoz/awesome-n8n-templates>
- File counts: the repository's own git tree, main, read 2026-09-05 (truncated: false, so the
census is complete rather than sampled).
- Stars, forks and last push: the GitHub API for that repository, read 2026-09-05.
- Licence: the API reports NOASSERTION — the repository is public and open to contributions, but it
does not resolve to a standard licence, so read its own terms before putting a workflow into
something commercial.