Univer Office SDK

Univer Office SDK is an open-source framework by DreamNum Co., Ltd. for building embeddable productivity experiences — spreadsheets, documents, presentations, Bases, Boards, and PDFs — inside your own product. It is licensed under Apache-2.0 and actively maintained on GitHub, with the latest release being v1.0.2 published in September 2026.

What It Is

Univer is a full-stack Office SDK built around a microkernel and plugin architecture, a Canvas2D rendering engine, and a unified Facade API that works identically in the browser and on Node.js. Rather than a hosted app or a fixed UI, it gives developers composable building blocks for spreadsheet, document, and presentation experiences. The SDK is designed so that both humans and AI agents can work in the same files through structured APIs, worktree collaboration, and programmatic editing.

Plugin Architecture and Composition Model

Every capability in Univer is delivered as a composable plugin. Developers can choose between two integration paths:

• Plugin Mode: manually register packages for strict control over dependencies, lazy loading, bundle size, and runtime composition.
• Preset Mode: use curated plugin collections (e.g., @univerjs/preset-sheets-core) for a working Sheets or Docs setup with minimal configuration.
• Headless Mode: exclude UI plugins entirely and run workbook/document processing in Node.js for server-side automation, formula calculation, and AI agent workflows.

The SDK ships 100+ official plugins covering sheets, docs, slides, collaboration, and more, and supports React, Vue, Angular, Next.js, Astro, Web Components, and plain JS integrations.

AI Agent and Headless Capabilities

The README describes Univer as "The Office Harness for AI Agents." The AI SDK enables agents to inspect and modify Office content through structured APIs, verify output through content inspection and rendered screenshots, and work in isolated worktree drafts that humans can review and merge. The headless runtime runs the same codebase in Node.js without a browser, making it suitable for LLM-driven document generation, server-side formula calculation, and automation pipelines. A companion repository, dream-num/univer-mcp, provides MCP integration for driving Univer Sheets with natural language.

Performance and Rendering

The SDK uses a Canvas2D rendering engine shared across document types. The homepage lists benchmark targets including loading 1 million cells, scrolling 6 million cells at 50–60 FPS, and calculating 20,000 formulas. The formula engine supports 500+ built-in functions, custom functions, array formulas, and named ranges. Import and export cover xlsx, csv, tsv, docx, pptx, and pdf formats (some via Univer Pro extensions).

Open Source and Pro Boundary

The core SDK, plugin system, rendering engine, formula engine, Facade API, themes, i18n, and framework adapters are all open source under Apache-2.0. Univer Pro is a separately developed commercial extension layer that adds real-time collaboration, edit history, import/export, printing, charts, pivot tables, server-side calculation, and enterprise deployment packages. The README explicitly states that OSS packages are intended to be useful on their own without requiring Univer Pro.

Update: v1.0.2

The latest release, v1.0.2 (published September 24, 2026), marks the first stable 1.x milestone for the SDK. The project has been under active development since 2022 and has accumulated over 17,900 GitHub stars and 1,500 forks as of the latest data. The repository is a TypeScript monorepo with packages for core engines, document types, UI plugins, and feature plugins, developed under the dev branch with continuous CI builds.