kove documentation

kove is a hosted document API. You describe the document in JSON, and kove returns a PDF that looks right. We run all the infrastructure: the rendering, the scaling, the Chromium. You run nothing.

The usual way to add documents to a product is slow: you fight with layout, CSS, page breaks, and infra just to produce a clean PDF. kove takes that off your plate.

Here is the part that matters most. You do not learn the API by hand. You tell your coding agent (like Claude Code) what you need, for example “add PDF invoices to my checkout”, and it wires kove’s API into your app: the call, the auth, the data. You review the change and ship it. Document generation becomes part of your product, and you barely notice there is a provider behind it.

curl -X POST https://api.kove.dev/v1/documents \
  -H "Authorization: Bearer $KOVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "document": { "body": [ { "type": "heading", "text": "Hello, kove" } ] } }'

Why kove

kove is declarative. You say what goes in the document: a header, a table, some totals. kove decides how it looks and lays out the pages. You never write a line of CSS.

That covers the parts that usually eat your time:

Tables that span many pages, with the header row repeated on each one.
Rows that never get cut in half across a page break.
A totals block that stays attached to its table.
Section titles that never get stranded alone at the bottom of a page.
Page breaks where you want them, not where the browser decides.

You get all of this from a plain JSON document.

Why your AI can wire it in so fast

kove is built to be easy for an AI to integrate. The model is a simple JSON document. There is a full OpenAPI spec at /openapi.json and an llms.txt for agents. So when you ask your coding agent to add documents to your app, it already has the whole contract in front of it: the endpoint, the auth, the document shape. It writes the integration, you review it, you ship.

How to use it

The hosted API is the main surface. It is what your AI integrates into your app, and what runs in production.

HTTP API · POST /v1/documents with your API key. Hosted render, zero infra. This is the primary way to use kove. See API.

Two more options sit alongside it:

CLI · npx kove render doc.json -o doc.pdf. Local render with Chromium, no account. Good for trying things out, scripts, and CI. See CLI.
MCP · for runtime agents that build or edit documents on the fly. Secondary, but handy when an agent needs to make a document live. See MCP.

All of them use the same document model. See Document model for the full list of building blocks.

Start here

Quickstart · try a render locally, then get an API key for your app.
HTTP API · the primary surface: auth, endpoints, errors, and limits.
Document model · the building blocks (table, totals, sections, and more).
CLI · render, login, keys, dev.
MCP · for runtime agents.
Auth and billing · account, API keys, free tier, and upgrade.