The fastest way to try DeepSeek Harness is a single npx command. What really matters is getting three things straight first: the prerequisites, where it runs, and how to configure it after launch. Nail those, and you'll be running within five minutes.

Prerequisites

• Node.js installed (^20.19 || >=22)
• Building from source also requires pnpm

Quick start via npm

On a system with Node.js installed, run:

npx @deepseek-ai/dsh web

This starts the Web UI, served at http://127.0.0.1:3080 by default. The command prints the actual address.

Build from source

Clone the full source and follow the repository instructions:

git clone https://github.com/deepseek-ai/deepseek-harness
cd deepseek-harness
pnpm install
pnpm run build
pnpm dsh web

After launch

Open the Web UI and do two things:

1. Configure a model: open Settings → Models, enter your DeepSeek API key and save. Model routing works immediately.
2. Pick a workspace: click Select workspace, add the project directory where you launched dsh, and select it (the session input stays disabled until you do).

After those two steps, you can run a task.

Want to see the config tree?

Run dsh --profile web --dump-config to see the actual configuration tree your machine boots — any entry it prints can be replaced by your own patch.