DeepSeek Harness Tutorial
DeepSeek HarnessDeveloped by DeepSeek AI and officially open-sourced in August 2026Agent Harness (Agent Framework), licensed under MIT, written in TypeScript.
DeepSeek Harness AdoptsEverything is a PluginOpen architecture: all Agent capabilities such as models, tools, skills, sessions, sandbox, storage, loops, scheduling, and UI are composed of plugins, and can be freely replaced and flexibly recombined—Customize your own Agent at the configuration layer without modifying the source code。
DeepSeek AI's official philosophy:Agent = Model + HarnessOfficial slogan:Everything is a Plugin. (Everything is a Plugin)
Who should read this tutorial?
This tutorial is intended for the following types of readers:
Developers / EngineersFor those who want a local AI coding assistant whose capability boundaries are fully controllable and every component can be replaced or recombined—rather than a black box locked by the vendor into fixed functionality.
AI/ML practitionersFor those interested in Agent architecture—the underlying Cordis plugin system drives it, the event-driven design is backed by academic papers, and the PTC pattern allows models to compose multi-step tool calls with TypeScript programs.
Efficiency tool enthusiasts: For those who wish to embed the Agent into their own workflow using a Web UI graphical interface, headless command line, or Python SDK.
Researchers: Need observable agents—every run writes to append-only session logs, with system prompts, chain-of-thought, tool calls, and sub-agent scheduling viewable in the Trajectory view.
Users who value open source and controllabilityMIT open-source license, free composition at the configuration layer, no telemetry, no cloud lock-in, fully self-controlled.
Prerequisite skills before reading
This tutorialNot requiredAn AI research background or in-depth machine learning knowledge. You need to have:
| Skills | Requirement level | Description |
|---|---|---|
| Command-line basics | Required | Ability to execute commands in a terminal and set environment variables |
| Node.js Basics | Required | Install and start dsh via npx / pnpm. |
| Basic API concepts | Familiarity is enough | Know what an API Key is and how to apply for a DeepSeek API Key |
| Plugin / configuration file concepts | Familiarity is enough | Used in advanced chapters (plugin development, configuration layer composition) |
| Git basics | Optional | Used when installing from source or contributing |
Operating system requirements: Linux, macOS, or Windows (can be run after installing Node.js).
Core features
DeepSeek Harness Features:
🧩 Everything is a plugin— The Cordis kernel only handles plugin loading, unloading, and dependency management; all capabilities such as models, tools, skills, sessions, sandbox, storage, loops, scheduling, and UI are provided by plugins, which collaborate through services and events, with free composition at the configuration layer
📜 Traceable execution— Everything the model sees (system prompts, chain-of-thought, tool calls and results, sub-agent scheduling, context injection) is written to append-only session logs; the Trajectory view displays by source, with resume, fork, retrieval, and replay sharing the same event stream
🛠️ Four operating modes— Standard mode (feature-complete coding Agent), PTC mode (the model combines multiple rounds of tool calls using a TypeScript program), Minimal mode (two tools, for model benchmarking), Creative mode (runtime checks + plugin experimentation + custom preset creation)
🌐 Multi-form usage— Web UI (default http://127.0.0.1:3080), headless (run once, print the final answer, and exit), CLI, Python SDK, TypeScript SDK
🔌 Rich tools and extensions— File editing, Shell, file and web search, Skills, planning, goals, subagents, workflows; event-driven extension points (session/Agent/capability three-level events), any capability can be replaced by patches
🔒 Open and controllable— MIT open source; Profile + composition package layered configuration,
dsh --profile web --dump-configView the complete configuration tree of the actual startup; unprivileged kernel, all registrations are reversible🤖 Model-agnostic— Fill in the DeepSeek API key and it's ready to use; supports other providers and custom OpenAI-compatible endpoints; model routing without restart
🔬 Research-ready— Fully observable event stream and configuration tree, open design for Agent architecture research (the underlying Cordis is backed by academic papers)
Related resources
Official resources
| Resources | Links |
|---|---|
| Official Website | deepseek.com/harness |
| GitHub repository | github.com/deepseek-ai/deepseek-harness |
| Official Docs · Quick Start | deepseek-harness.github.io/deepseek-harness/guide/quickstart |
| npm package | @deepseek-ai/dsh（npx @deepseek-ai/dsh webone-click startup) |
| Plugin Community | GitHub topics: dsh-plugin |
| Community Support | GitHub Discussions |