Auto-caption your posts with n8n + Caption-Pack (5 minutes, free)

Last week I launched a small API that turns a topic, audience, and tone into ready-to-post social captions — no LLM in the loop, just a deterministic template-and-variant engine. It works. But an API only helps people who wire it up, so today I'm sharing the easiest on-ramp I know: an 8-node n8n workflow that turns one piece of content into platform-ready posts for X, LinkedIn, and Instagram in about five minutes.

If you've never touched n8n, this is a friendly first workflow. If you live in n8n, steal it and make it yours.

What you'll build

A form where you paste your source content (an announcement, a blog excerpt, a key message), pick your platforms and tone — and get back a complete multi-platform content plan: captions + hashtags for each platform, each in a platform-appropriate tone (X → bold, LinkedIn → professional, Instagram → warm — or your own pick), plus a suggested posting order and your remaining credit balance. Each platform costs 1 API credit per run.

Step 1: Get a free API key

Open this in your browser:

https://caption-pack-api.onrender.com/v1/free-trial

It hands you a key with 25 free calls, no card. The key is shown once, so save it somewhere safe (a password manager works fine).

Step 2: Import the workflow

Grab the template JSON from the repo:

https://github.com/jeshuadomingo-byte/caption-pack-n8n-template

In n8n: Workflows → ⋯ → Import from File, and select caption-pack-n8n-template.json. You'll see eight nodes plus friendly sticky notes:

• ContentForm — the input form (source content, audience, platforms, tone, count per platform)
• Build Platform Jobs — a Code node that builds one API job per platform with per-platform tone defaults
• Loop Over Platforms — runs the API call once per platform
• Call Caption-Pack API — the HTTP request
• Collect Packs — gathers each platform's caption pack
• Check Balance — reads your remaining credits
• Build Content Plan — compiles everything into one markdown posting plan
• Show Content Plan — the completion page

Step 3: Connect your key

n8n strips credentials on export (by design — never share keys in a template), so you wire yours in once:

1. In n8n, go to Credentials → New → Header Auth.
2. Set Name to Authorization and Value to Bearer followed by your key (one space after Bearer).
3. Open the Call Caption-Pack API node and the Check Balance node, and select that credential in both.

Step 4: Run it

Click Test workflow, fill in the form (try pasting a short announcement as the source, audience "remote workers", platforms "All three", tone "Auto"), and submit. Your multi-platform content plan lands on the completion page — per-platform captions with hashtags, a suggested posting order, and your credit balance underneath.

That's it. Five minutes, zero dollars.

Make it yours

The Form trigger is the friendliest demo, but the real fun starts when you swap it:

• Schedule trigger + Google Sheets — generate a fresh batch of caption ideas every morning and append them to a sheet. Hands-off content pipeline.
• Webhook trigger — POST a JSON body like {"topic": "...", "audience": "...", "tone": "bold", "platform": "x", "count": 5} from your own app or AI agent, and get captions back. This is the lane I'm most excited about: agents calling a deterministic caption engine instead of burning LLM tokens.

The cost math

1 credit per platform per run (3 for all three). $10 gets you 500 calls — that's 2¢ a caption batch. The free tier covers this entire tutorial with room to spare.

Troubleshooting

Why no LLM?

Deterministic means predictable: same inputs, same quality, every time — and at 2¢ a run, it's cheaper than any model call that could do the job. There's a time for LLMs; caption drafting at volume isn't it.

The template is MIT-licensed. If you build something with it, I'd genuinely love to hear about it — that's what keeps a project like this alive.

Full API contract: https://caption-pack-api.onrender.com/openapi.yaml
Try the API with no key at all: the keyless playground is on the landing page.