github.devotd.
NewAI

dream-num/univer-workspace agent-v0.1.0 — Agent Office Workspace

dream-num/univer-workspace is an Apache-2.0 open-source office workspace where people and AI agents create and review documents, sheets, and slides together; the latest release is agent-v0.1.0.

Repositorydream-num/univer-workspace ↗

dream-num/univer-workspace is an Apache-2.0 open-source office workspace where people and AI agents create and review documents, sheets, and slides together; the latest release is agent-v0.1.0.

GitHub description: An open-source Office workspace where people and AI agents create, collaborate, and review together.

stars
★ 2.3K
Forks
221
Language
TypeScript
License
Apache-2.0
Latest release
agent-v0.1.0
Read original source ↗View on GitHub ↗

dream-num/univer-workspace is an open-source office workspace where people and AI agents create and review documents, sheets, and slides together. The latest public release is agent-v0.1.0, and the README describes the repository as a server, browser, and agent configuration. What this release changed is not included in the information given.

It handles spreadsheets, documents, presentations, relational tables, and boards in a single collaborative space, lets you edit documents in the browser, and leaves document chat and change review to the agent app, while automation is handled through a command-line interface. Agents are designed to modify documents in a separate workspace, verify them, and then hand them to a human for review and merge. The repository contains server and web app, agent app, and desktop app implementations; the primary language is TypeScript and the license is Apache-2.0. The GitHub star count is 2263.

The center of this architecture is that changes made by agents are not immediately applied to the source. Review and merge remain with people, and the browser, the agent app, and the command line share the same data and permission model. What distinguishes this repository is the setup that bundles the document editor and the agent workspace into a single server. However, the repository was created on August 14, 2026.

The research record scores 8.0 for technology, 7.0 for quality, 6.0 for novelty, and 0.0 for momentum, out of 10. There are no figures for performance, concurrency, or the accuracy of agent changes. What is being examined here is the direction of this architecture, not a case that demonstrates the reliability of real document work.