Skip to content

AgentsDeepSeek

DeepSeek open-sources DeepSeek Harness, a plugin-based coding agent, alongside the V4-Pro general release

DeepSeek Harness at a glance

DeepSeek Harness (dsh) is an MIT-licensed agent harness in developer preview, built so that models, tools, sessions, sandboxes and interfaces are all swappable plugins. It runs as a local web app and works with DeepSeek, Anthropic, OpenAI and custom endpoints.

What changed with DeepSeek Harness

On August 13, 2026, DeepSeek made DeepSeek-V4-Pro generally available on its web app, mobile app and API, and the same day published DeepSeek Harness on GitHub under the MIT license. The project calls itself an open-source agent harness; VentureBeat framed it as an open rival to integrated coding-agent tools such as Claude Code.

Harness is built on Cordis, a plugin framework, around the principle that everything is a plugin: models, tools, skills, sessions, sandboxes and user interfaces can each be replaced. Running npx @deepseek-ai/dsh web starts a local web interface at 127.0.0.1, and VentureBeat reported a headless mode and a Python SDK. It can inspect and edit repositories, run shell commands, search, plan, spawn subagents and apply configurable approval policies.

It is not limited to DeepSeek models; Anthropic, OpenAI and custom compatible endpoints also work. DeepSeek labels it a developer preview and warns in capital letters that there will be compatibility-breaking changes, and it asks users to read a safety notice before running it. By November 5, 2026, the repository had passed 270,000 GitHub stars.

DeepSeek Harness in numbers

Values are copied from the linked source on the date shown. They are not IA Compa measurements.

GitHub stars (deepseek-ai/deepseek-harness)as of · Source for GitHub stars (deepseek-ai/deepseek-harness): github.com (opens in a new tab)
273,980

Where to compare DeepSeek Harness

Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.

Why DeepSeek Harness matters for businesses

For businesses, a major lab releasing its agent harness as open source means more choice in how coding agents are assembled and which models sit behind them. Because it is a fast-moving preview, pin versions, run it in a sandbox, and keep it away from production credentials until the interfaces settle.

Put it to work: AI Agents

Sources for this story

  1. Official
  2. deepseek-ai/deepseek-harness (archived August 13, 2026) (opens in a new tab)

    DeepSeek on GitHub via Internet Archive · web.archive.org

    Official
  3. DeepSeek-V4-Pro GA Release (opens in a new tab)

    DeepSeek API Docs · api-docs.deepseek.com

    Official
  4. Press

Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.

  • #agents
  • #coding-agents
  • #deepseek
  • #open-source
  • #agent-harness

Want this working in your business?

We turn new AI capabilities into agents, voice receptionists, and automations you can actually use.