TL;DR
Get tech for your team delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
DeepSeek has launched DeepSeek Harness in a worldwide public preview as an open-source AI agent platform. The company describes it as available as a desktop app or through a web interface, with plugins for tasks including document work, coding, research and scheduled jobs; download details and supported desktop operating systems are not specified in the supplied announcement.
DeepSeek has put DeepSeek Harness, an open-source AI agent platform, into public preview worldwide, describing it as usable as a desktop app or through a web interface launched from code. The release matters to developers and other users interested in running AI-assisted workflows because DeepSeek says Harness can be extended with plugins for tasks such as coding, document work, research and scheduled jobs.
DeepSeek presents Harness as a tool for everyday work and software development. Its examples include organizing files, analyzing data, drafting documents and creating slides, alongside exploring code repositories, fixing bugs, building features and running tests. The page also lists research tasks, including finding information, checking facts and citing sources. These are descriptions of intended capabilities from DeepSeek, not independent evaluations of how well the software performs.
The product is built on what DeepSeek calls Cordis’s “everything is a plugin” architecture. According to the company, users can install plugins or create them through conversation in “Creator mode” to add tools, skills and interface features. DeepSeek also describes a Scheduled tasks plugin for starting work at a chosen time or running recurring tasks, plus developer tools that expose execution traces and runtime information.
For developers, the announcement gives a command to start the web interface: npx @deepseek-ai/dsh web, after installing Node.js. It also points users to a GitHub repository for cloning the source and following setup instructions. The supplied page refers to a desktop app but does not provide specific installation links, system requirements or a supported operating-system list.
A Plugin Platform for Agent Work
Harness gives DeepSeek a way to present its models as part of a customizable working environment, rather than only as a chat interface. If its plugins work as described, users could connect an agent to a wider range of tasks and tailor its tools or interface to particular workflows. Open-source availability may also let developers inspect and adapt the project, although the announcement alone does not establish the maturity or limits of that code.
The desktop framing could make agent workflows more accessible to people who prefer an installed application, while the web launch command offers a route for technical users. Those are meaningful distinctions for prospective users deciding whether to try the preview. However, DeepSeek has not provided performance results, independent security assessments or detailed information about data handling in the supplied material, so the announcement should be read as a product release description rather than proof of capability or safety.
As an affiliate, we earn on qualifying purchases.
From Source Code to Preview
DeepSeek’s page describes Harness as open source and in public preview, and offers two ways to get started: a desktop application or a web interface launched with Node.js. It directs developers to the project repository at github.com/deepseek-ai/deepseek-harness and gives the npm command for the web version. The announcement does not provide a detailed release history or explain how the preview differs from earlier internal or developer builds.
The product page also names DeepSeek-V4.1-Flash in its interface preview. That mention is not accompanied by model specifications, pricing, availability terms or an explanation of whether users can select other models. The page’s emphasis is instead on the Harness layer: the environment for issuing tasks, using plugins and inspecting tool activity.
““DeepSeek Harness is now in public preview worldwide and open source, with composable plugins that extend what agents can do.””
— DeepSeek, on the Harness product page
As an affiliate, we earn on qualifying purchases.
Preview Details Still Missing
The supplied announcement does not confirm which macOS and Windows versions are supported, whether desktop builds are available for download now, or how installation works. It also gives no system requirements, licensing terms beyond the open-source description, or information about regional restrictions and pricing. The phrase “worldwide” signals the stated reach of the public preview, but the page does not explain whether every feature is available in every location.
Other open questions include how Harness handles user files and task data, what permissions plugins can access, which models are supported, and whether scheduled tasks continue running when the app is closed. DeepSeek lists possible uses and developer features, but provides no independent testing or benchmarks in the supplied source. Those details will matter to users weighing the preview for sensitive work or recurring automation.
As an affiliate, we earn on qualifying purchases.
Installation and Usage Details
Readers seeking to try the web interface can follow DeepSeek’s stated route: install Node.js, then run npx @deepseek-ai/dsh web. Developers can consult the linked GitHub repository for source and setup guidance. For desktop users, the next practical step is to check DeepSeek’s Harness page or repository for macOS and Windows download instructions and compatibility details, which are not included in the supplied announcement.
As the preview develops, the information most useful to users will include desktop release links, supported systems, data and plugin permission policies, and any changes to the feature set. DeepSeek has not provided a timetable for a stable release or named a next update in the source material.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is DeepSeek Harness?
DeepSeek describes Harness as an open-source AI agent platform for tasks including document and data work, coding, research and background jobs. It is designed to be extended through plugins.
Is DeepSeek Harness available for macOS and Windows?
The announcement describes a desktop app, but the supplied material does not specify supported operating systems or provide separate macOS and Windows download links. Check DeepSeek’s product page or repository for current installation details.
How can developers launch the web interface?
DeepSeek says to install Node.js and run npx @deepseek-ai/dsh web. Its page also links to the source repository for setup instructions.
How do plugins work in Harness?
According to DeepSeek, users can install plugins or create them through chat in Creator mode. Plugins can add tools, skills and interface features, though the announcement does not detail permissions or security controls.
Is the preview a stable release?
No stable-release date is given. DeepSeek labels Harness a public preview, and the supplied announcement does not say when that phase will end or what features may change.
Source: hn
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
