okraPDF

okraPDF turns a PDF into agent-ready data: clean markdown, extracted tables and figures, and grounded answers with page-and-bbox citations. Everything is one set of resources — documents, files, jobs, collections — reachable from whichever surface fits your workflow.

npm install -g @okrapdf/cli
okra upload report.pdf
okra chat "What was Q3 revenue?" --doc <docId>

Pick your surface

CLI

The agent-first okra command. Upload, extract, chat, and query collections from a script or an agent loop.

API

REST endpoints under api.okrapdf.com, plus the TypeScript SDK and React hooks.

MCP

Connect okraPDF to Claude, Cursor, or any MCP client and read PDFs in chat.

Web

The visual workspace at app.okrapdf.com — the same resources, in a browser.

iOS

A native PDF toolkit for iPhone and iPad, backed by your okraPDF account.

One resource graph, every surface

A PDF you upload with okra upload is the same document you open on app.okrapdf.com, read over the API, or query through MCP. The nouns never change:

Resource	What it is
`documents`	Processed PDFs with markdown, page images, tables, chat, and citations
`files`	Raw uploaded assets not yet turned into a document
`jobs`	Parse, render, and workflow jobs with status and results
`collections`	Groups of documents you can query together

Start with the CLI

Install @okrapdf/cli and extract your first PDF in under a minute.

Getting Started

Documents

Jobs

Collections

Reference

Pick your surface

CLI

API

MCP

Web

iOS

One resource graph, every surface

Start with the CLI

​Pick your surface

CLI

API

MCP

Web

iOS

​One resource graph, every surface

Start with the CLI

Pick your surface

One resource graph, every surface