DocMint

POST your JSON against a Word, Excel or PowerPoint template. Get the Office file, a PDF, or both.

DocMint fills .docx, .xlsx and .pptx templates from a workflow — no Office install, no COM automation, no per-format library to license. It will also tell you what fields a template needs before you send any data, and it refuses to render a document with an unresolved placeholder instead of shipping one with a hole in it.

There is a hosted service at docmint.app.mintapis.com — 30 documents a month free, no card — and an n8n community node that turns a template's placeholders into real n8n fields.

This repository is the API behind both.

Make your first document

Create a free account, copy its API key, and run this from a clone of this repository (git clone https://github.com/fstandhartinger/docmint.git && cd docmint). You need Node.js 20+ and curl, but no npm install, Office installation or template upload. The request uses the existing sales-report example and its complete sample data, rather than a template name your new account does not have yet.

export DOCMINT_API_KEY='paste-your-api-key-here'
node -e 'const fs=require("node:fs"); process.stdout.write(JSON.stringify({template_base64:fs.readFileSync("examples/sales-report.xlsx").toString("base64"),data:JSON.parse(fs.readFileSync("examples/sales-report.data.json","utf8")),output:"document"}))' > request.json
curl --fail-with-body --max-time 120 \
  https://docmint.app.mintapis.com/v1/render \
  -H "Authorization: Bearer $DOCMINT_API_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @request.json \
  --output filled-sales-report.xlsx

Open filled-sales-report.xlsx: it contains the filled sales workbook, including the regional summary. This costs one of your 30 free credits. For a PDF, change output:"document" to output:"pdf" and the output filename to filled-sales-report.pdf; that costs two credits. If curl reports an HTTP error, the output file contains the API's JSON explanation, not an Office file. Check your remaining credits with:

curl --fail-with-body https://docmint.app.mintapis.com/v1/usage \
  -H "Authorization: Bearer $DOCMINT_API_KEY"

For n8n, use the existing DocMint node setup guide instead; you do not need to write the HTTP request yourself.

What it does

Endpoint	What it is for
POST /v1/render	Template + data → the Office file, a PDF, or both
POST /v1/render/batch	Many documents in one call
POST /v1/inspect	Upload a template, get its fields back — no render, no credit
GET /v1/templates/:name/fields	The same, for a template already stored
POST /v1/templates	Store a template; versions and rollback included
POST /v1/jobs	The same renders, asynchronously, with a webhook
GET /v1/capabilities	What this deployment can actually do, from the code
GET /v1/usage	Plan, credits used, credits left

One placeholder syntax across all three formats, images included: {{customer.name}}, loops, conditionals and table-row expansion behave the same in a Word document, a spreadsheet and a slide deck.

The fill engine is ours. SPEC.md explains why: the obvious library covers DOCX and PPTX for free but hard-codes XLSX behind a paid module, along with images and HTML-into-DOCX. Writing the OOXML work ourselves meant one syntax everywhere, no paid ladder later, and full control over the behaviour this product is judged on — failing loudly on an unresolved placeholder, by name and by location.

LibreOffice (MPL-2.0, from the Debian archive) does the PDF conversion.

The full reference is at docmint.app.mintapis.com/docs.

Plans

Plan	Documents / month	Price
Free	30	$0
Starter	2,000	$9
Pro	20,000	$29
Scale	100,000	$99

A PDF conversion costs a credit on top of the fill, because it costs a LibreOffice process. The docs say so on the pricing section rather than in a footnote.

What it costs to run

The canonical hosted service runs on Sandy via Coolify; see DEPLOY.md for its repository, branch and deployment process. A legacy Render endpoint is also retained for existing integrations. The old Render/Neon inventory in ops/INFRASTRUCTURE.md is historical, not a current total-cost estimate or a safe teardown checklist for the hosted service.

LibreOffice peaks at 219 MB per conversion, measured — which is why MAX_CONCURRENT_PDF defaults to 1 in a 512 MB container.

Running it yourself

A plain Express app, a Postgres database, and soffice on the PATH for PDF output. The Dockerfile installs libreoffice-writer, libreoffice-calc and libreoffice-impress, one per input format: the .xlsx import and calc_pdf_Export ship in the Calc package and the .pptx ones in Impress, so a host with Writer alone can turn only Word templates into PDFs.

npm install
cp .env.example .env.local     # DATABASE_URL and SESSION_SECRET are the only ones needed
npm run migrate
npm start                      # http://localhost:3000
npm test                       # needs a database; Stripe tests skip without keys

Without soffice, everything still works except "output":"pdf", and GET /v1/capabilities says so rather than failing at render time.

Related

• n8n-nodes-docmint — the n8n community node. It reads your template and builds one typed, expression-capable n8n field per placeholder, named from the document itself.
• PDFMint — the same idea for HTML, Markdown and URLs.

License and status

Run by one person. Issues and pull requests are read. If a document comes out wrong, an issue with the id from the X-DocMint-Request-Id header is enough to find it in the logs.